--- Log opened Mon Oct 10 00:00:57 2016
20161010 00:08:33-!- tad_ [~lundberg@173.217.65.103] has quit [Quit: Leaving]
20161010 00:23:31-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has quit [Remote host closed the connection]
20161010 00:23:40-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has joined #wesnoth-dev
20161010 00:31:28< vultraz> gfgtdf: DeFender1031, yeah, the number of colors in mp create is hardcoded to max 10  
20161010 00:31:39< vultraz> return custom_color_.empty() ? gamemap::MAX_PLAYERS : gamemap::MAX_PLAYERS + 1;
20161010 00:31:56< vultraz> er, mp connect (gui1) or mp staging (gui2)
20161010 00:31:59< vultraz> will have to fix this
20161010 00:33:20< vultraz> and yes, color handling needs to be cleaned up :|
20161010 00:33:24< vultraz> some places use Uint32
20161010 00:33:29< vultraz> some places use SDL_Color..
20161010 00:33:54< vultraz> some places use the game_config colors by index or name..
20161010 00:34:12< vultraz> as well as various functions scattered everywhere to get/convert between any of those
20161010 00:34:32< Aginor> I started to look into that
20161010 00:34:48< Aginor> and we've also have code that's using ARGB and RGBA, depending on where
20161010 00:34:55< vultraz> Oh god :|
20161010 00:34:56< Aginor> it's messy
20161010 00:34:57< vultraz> whyyy
20161010 00:35:08< Aginor> pango
20161010 00:35:20< Aginor> but it seems to have been brought across into multiple places
20161010 00:35:27< vultraz> what about pango?
20161010 00:35:35< vultraz> pango uses hex color.
20161010 00:35:39< Aginor> it only supports RGBA
20161010 00:35:45< Aginor> or cairo even
20161010 00:35:47< vultraz> right
20161010 00:35:49< Aginor> same difference
20161010 00:35:55< vultraz> where would we need argb?
20161010 00:36:07< Aginor> that's what we do for all rendering code
20161010 00:36:14< vultraz> uh...
20161010 00:36:16< vultraz> no?
20161010 00:36:20< vultraz> GUI2 uses RBGA 
20161010 00:36:28< Aginor> alpha mask is 0xff000000
20161010 00:36:32< Aginor> that's argb
20161010 00:36:42< Aginor> if it was 0x000000ff it'd be rgba
20161010 00:36:51 * vultraz groans
20161010 00:37:04< Aginor> I've started to tidy it up a bit ot make it clearer, but I ran out of time
20161010 00:37:13< vultraz> I can imagine.
20161010 00:37:27< vultraz> Did you make any progress with the custom blitting stuff, btw?
20161010 00:37:29< Aginor> and now I need to disappear
20161010 00:37:36< Aginor> it was all intermingled
20161010 00:37:54< Aginor> I know kind of what's going on, but not how to reproduce it without that blitting code
20161010 00:37:59< Aginor> but it'd be good to get rid of it
20161010 00:38:07< Aginor> (very good)
20161010 00:38:55< vultraz> Isn't it necessary?
20161010 00:38:58< EliDupree> Retrieving persistent variables of sides that aren't the current player: does it work in 1.12? I remember it's been fixed, but I don't remember if the fix is in 1.12 branch or just 1.13
20161010 00:39:03< vultraz> To be gotten rid of, that is
20161010 00:39:46-!- irker693 [~irker@uruz.ai0867.net] has quit [Quit: transmission timeout]
20161010 00:47:01< EliDupree> Found it in the change log – looks like this is the fix, in 1.11.13
20161010 00:47:02< EliDupree> * [get_global_variable] now works properly when side != current_side.
20161010 00:55:04< EliDupree> Now I'll have to do loads of work to disentangle my code from the complex workarounds I created to work around that bug, LOL.
20161010 01:06:23-!- gfgtdf_ [~chatzilla@x4e36a1e7.dyn.telefonica.de] has joined #wesnoth-dev
20161010 01:07:22-!- tad_ [~lundberg@173.217.65.103] has joined #wesnoth-dev
20161010 01:08:44-!- gfgtdf [~chatzilla@x4e363dc4.dyn.telefonica.de] has quit [Ping timeout: 260 seconds]
20161010 01:08:50-!- gfgtdf_ is now known as gfgtdf
20161010 01:15:36< celticminstrel> vultraz: Perhaps a map<string, node_path> instead of a map<string, node*?> Though that'd have to be partially rebuilt anytime the teams change.
20161010 01:15:37< celticminstrel> gfgtdf: We have more than 9 team colours if you count the orb options. (Though those probably aren't exposed as team colours in the UI.)
20161010 01:15:57< vultraz> celticminstrel: cannot talk, watching the Debate
20161010 01:17:24< celticminstrel> Yay debates.
20161010 01:17:37 * celticminstrel is playing a game, so wouldn't be very responsive anyway,
20161010 01:18:09< tad_> Probably most of the country is watching.
20161010 01:18:18< celticminstrel> Which of course excludes me. :P
20161010 01:18:38-!- ancestral [~ancestral@75-168-189-115.mpls.qwest.net] has joined #wesnoth-dev
20161010 01:18:41< celticminstrel> Oh hey, zookeeper changed UtBS to use type_tree.
20161010 01:21:14< tad_> gui/widgets/listbox.cpp:334:73: warning: ‘*((void*)& horizontal_scrollbar_position +4)’ may be used uninitialized in this function
20161010 01:29:41-!- Rh0nda [~rhonda@anguilla.debian.or.at] has quit [Ping timeout: 248 seconds]
20161010 01:30:29-!- Jetrel_bot [~Jetrel@ec2.happyspork.com] has quit [Ping timeout: 260 seconds]
20161010 01:31:13-!- Rhonda [~rhonda@anguilla.debian.or.at] has joined #wesnoth-dev
20161010 01:34:39-!- Jetrel_bot [~Jetrel@ec2.happyspork.com] has joined #wesnoth-dev
20161010 01:34:46-!- Grickit is now known as Gambit
20161010 01:35:21< shadowm> tad_: Is that on master?
20161010 01:36:23< tad_> yes. I didn't notice if it showed using cmake but I'm using scons and it showed
20161010 01:36:43< shadowm> Didn't for me, probably gcc version difference.
20161010 01:37:06< shadowm> (At least the wording sounds like gcc to me. I'm assuming here.)
20161010 01:37:26< tad_> I did a CMake to check for problems and didn't see any, then immediately starter scons.  yes, gcc
20161010 01:38:22< tad_> $ gcc --version
20161010 01:38:22< tad_> gcc (GCC) 6.2.1 20160830
20161010 01:38:39< shadowm> Yeah, I've not installed that yet, still on 6.1.1.
20161010 01:38:57< shadowm> Why isn't it warning about line 330 instead?
20161010 01:39:28< shadowm> Oh, right, it's the field it wants, optional isn't a POD type.
20161010 01:39:51< tad_> There were two. 334 and 333 
20161010 01:39:53< shadowm> Seems false positive to me then.
20161010 01:40:13< tad_> Probably but should be correctible nonetheless
20161010 01:40:53< shadowm> The bool conversion is supposed to yield true if there's a value, therefore the value is initialized.
20161010 01:41:04< shadowm> That's the only way to branch into that code.
20161010 01:45:02< tad_> I've not looked at the code. scons is still rebulding all and the debate is on
20161010 01:47:08< tad_> warning: ‘FriBidiLevel fribidi_log2vis(const FriBidiChar*, FriBidiStrIndex, FriBidiParType*, FriBidiChar*, FriBidiStrIndex*, FriBidiStrIndex*, FriBidiLevel*)’ is deprecated
20161010 01:48:16< tad_> font.cpp:631:85: warning: ignoring return value of ‘FriBidiLevel fribidi_log2vis(const FriBidiChar*, FriBidiStrIndex, FriBidiParType*, FriBidiChar*, FriBidiStrIndex*, FriBidiStrIndex*, FriBidiLevel*)’, declared with attribute warn_unused_result
20161010 01:48:59< shadowm> src/campaign_server/fs_commit.cpp:48:22: error: invalid pure specifier (only ‘= 0’ is allowed) before ‘::’ token
20161010 01:49:02< shadowm>   const int errno = std::errno;
20161010 01:49:07 * shadowm facekeyboards.
20161010 01:52:09< shadowm> I add one file to the campaignd target and scons suddenly needs to rebuild everything?
20161010 01:52:20< shadowm> Oh right, I forgot I modified the filesystem API.
20161010 01:53:28< shadowm> It's not everything anyway.
20161010 01:54:31< shadowm> Well, my code runs.
20161010 01:54:35< tad_> First time ever for me, so it is for me. Which may explain why the errors are suddenly showing to me
20161010 01:54:57< shadowm> I don't think I'll find out if it fixes what it needs to fix though.
20161010 01:55:42< shadowm> Oh, I know, I could tell campaignd to SIGKILL itself in one of the problematic sequences.
20161010 01:57:15< shadowm> Hmph. http://pastebin.com/raw/vZMdUC9F
20161010 01:58:57< shadowm> Well, at least now server.cfg can be unwritable and that doesn't matter if it can be deleted.
20161010 02:01:47< shadowm> Okay, it does what it's supposed to do. No surprise there.
20161010 02:04:40< shadowm> So I'm going to PR this and see what you people think.
20161010 02:30:32< shadowm> I think I made GitHub sad.
20161010 02:31:36< tad_> Oh oh. Using scons: touch src/lua/lvm.h; scons --> all targets are up to date. Doesn't scons use makedepend or something like that to rebuild when a #include changes?
20161010 02:34:02-!- gfgtdf [~chatzilla@x4e36a1e7.dyn.telefonica.de] has quit [Quit: ChatZilla 0.9.92 [Firefox 49.0.1/20160922113459]]
20161010 02:34:55< tad_> Oh, nevermind. I made an actual change and it spotted it.
20161010 02:45:45< shadowm> It uses checksumming IIRC.
20161010 02:48:52-!- esr [~esr@wesnoth/developer/esr] has quit [Quit: WeeChat 1.4]
20161010 02:49:43< shadowm> Okay, PR is up: https://github.com/wesnoth/wesnoth/pull/818
20161010 02:50:03< shadowm> And yes I know there's an unused forward decl of config. I'm just as sad about it.
20161010 02:50:52-!- wedge009 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has quit [Ping timeout: 260 seconds]
20161010 02:59:40< tad_> I have scons changed to inject the include and dependency for Lua and it works. But I see a deprecated function warning from gcc about using tmpnam instead of mkstemp .. did I cause that or is it known?
20161010 03:05:49< tad_> I rolled back to master and it went away. It must have come from building the objects as targets (required for my work) instead of implicitly
20161010 03:06:39< tad_> Well, there's a compiler flag for that. And for a couple other issues, so I guess I'll fix it with subject commits for each issue.
20161010 03:28:10< Aginor> 13:38 < vultraz> Isn't it necessary?
20161010 03:28:18< Aginor> I was referring to our custom blit function
20161010 03:31:00< vultraz> [11:38:58]	vultraz	To be gotten rid of, that is
20161010 03:35:26< Aginor> vultraz: only if you want to move to hardware acceleration
20161010 03:35:28< Aginor> otherwise, no
20161010 03:35:35< vultraz> But we do.
20161010 03:35:47< Aginor> having read the SDL blitting code, I think theirs might be faster too
20161010 03:36:03< vultraz> than?
20161010 03:36:20< Aginor> blit_surface
20161010 03:36:42< vultraz> of course.
20161010 03:37:21-!- irker158 [~irker@uruz.ai0867.net] has joined #wesnoth-dev
20161010 03:37:21< irker158> wesnoth: Charles Dang wesnoth:master b662e03dadd9 / data/gui/widget/window_tooltip_large.cfg: Tooltips: fixed a tiny shading inconsistency https://github.com/wesnoth/wesnoth/commit/b662e03dadd916ee57d9514f629992a207e1763c
20161010 03:37:33< vultraz> The bar for 'faster than our implementation' is very low in general.
20161010 03:38:08< vultraz> but the question here is whether you can make sdl_blit work right with text
20161010 03:38:22< vultraz> and images
20161010 03:38:29< vultraz> but mostly text
20161010 03:39:31< vultraz> and you seem to not be able to say anything about that
20161010 03:40:15-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Quit: Konversation terminated!]
20161010 03:53:52-!- Kwandulin [~Miranda@p200300760F2C71D3D56812C2C9812869.dip0.t-ipconnect.de] has joined #wesnoth-dev
20161010 03:55:59-!- wedge009 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has joined #wesnoth-dev
20161010 03:57:36< wedge009> tad_: What question?
20161010 03:58:48< tad_> Oh, I forget what I wanted then, but probably I updated my thinking. IIRC you use VS 2013?
20161010 04:03:15-!- wedge009 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has quit [Quit: wedge009]
20161010 04:03:34-!- wedge009 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has joined #wesnoth-dev
20161010 04:03:48-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has quit [Remote host closed the connection]
20161010 04:03:51< wedge009> VC14 / VS 2015
20161010 04:03:56< shadowm> Who is GregoryLundberg?
20161010 04:04:02< tad_> Me
20161010 04:04:14-!- vultraz_iOS [uid24821@wesnoth/developer/vultraz] has joined #wesnoth-dev
20161010 04:05:23< shadowm> Huh.
20161010 04:05:30< shadowm> I assumed your forum username was your real name.
20161010 04:06:49< tad_> wedge009, OK. Well I'm doing a changeset which involves VS project changes. I don't know how they've been done inthe past, but I did a diff and included it in my branch. Was thinking you might vet it in case I'm doing it wrong.
20161010 04:07:31< tad_> shadowm, At 'game' level I use one of two alias online. For 'real' work like github I use my real name.
20161010 04:19:57-!- bumbadadabum [~bumbadada@wesnoth/developer/bumbadadabum] has quit [Ping timeout: 240 seconds]
20161010 04:20:16-!- ancestral [~ancestral@75-168-189-115.mpls.qwest.net] has quit [Quit: i go nstuf kthxbai]
20161010 04:21:57-!- ChipmunkV [~vova@static-89-94-113-91.axione.abo.bbox.fr] has joined #wesnoth-dev
20161010 04:25:26-!- ChipmunkV [~vova@static-89-94-113-91.axione.abo.bbox.fr] has quit [Client Quit]
20161010 04:30:03-!- wedge009 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has quit [Quit: wedge009]
20161010 04:30:18-!- wedge009 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has joined #wesnoth-dev
20161010 04:33:17< wedge009> tad_: I'm no expert but I can have a look at what you've done.
20161010 04:37:04< tad_> wedge009, OK. My personal fork: GregoryLundherg/GL_Lua2 .. I've tested on Visual Studio Community 2015, and Arch Linux CMake and Scons.
20161010 04:37:40< tad_> grr typod my own name s/h/b/
20161010 04:39:34< wedge009> Heh, I got it.
20161010 04:45:34< wedge009> tad_: Am I looking at the wrong place? I'm going through https://github.com/GregoryLundberg/wesnoth/commit/0bae7dccccb349329c174c06e1205ea06020ac82
20161010 04:46:30< tad_> That is correct.
20161010 04:47:08< wedge009> Your commentary says it's already working for VS, right? What's to check?
20161010 04:49:23< tad_> wedge009, My concern is that I don't know what to do to backport 2015 to 2013 for the projectfiles. So this is my 'seems right' best guess. I was thinking you were on 2013 and could try it. I suppose I could install it on an 8.1 VM to test.
20161010 04:51:07< tad_> Right now I'm looking at side installing CodeBlocks to check it for MSVC and MinGW (assuming that's why we have two Codeblocks). I can't do a think about XCode, though.
20161010 04:52:05< wedge009> I don't see any changes for VS and I don't think there's that much difference between VC12 and VC14. If it works without any changes for VC14, I'm sure it'll work fine for VC12. If not, those who are still using it are sure to pipe up about it.
20161010 04:52:24< wedge009> Interesting, I just noticed there's no such thing as VC13. They must be superstitious.
20161010 04:53:06< wedge009> *pipe up about it... if there are any issues.
20161010 04:53:24< tad_> MS does an even-odd cycle. Odd years are Visual Studio, even years are the compiler. You _can_ install the compiler (it's free) witjout VS if you want to do only makefile builds.
20161010 04:54:29< tad_> That is why CodeBlocks can compile using Visual C/C++ compiler without having to install Visual Studio.
20161010 04:54:42< wedge009> Huh, I was looking at https://en.wikipedia.org/wiki/Microsoft_Visual_Studio#History
20161010 04:55:03< wedge009> Internally the VS <year> editions have their own number.
20161010 04:55:23< wedge009> 5, 6, 7, 8, 9, 10, 11, 12, 14, 15 - no 13.
20161010 04:55:37 * tad_ chuckles.
20161010 04:57:41< tad_> Anyway .. VS 2015 installed MinGW for me .. I need to check if it's a full install or if it's missing parts. No sense having two if I can avoid it.
20161010 05:00:40< wedge009> I don't remember that option. It might be customised for VS anyway.
20161010 05:01:03-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has joined #wesnoth-dev
20161010 05:02:27< tad_> Looked pretty stock to me but I didn't dig too deep. It might be a sub-set and I probably can fix that. There's really not a lot to MinGW and IIRC it was pretty alacarte.
20161010 05:04:47< tad_> Sorta strange though. I have a Windows box which can run GNU cross compiled (MinGW) and native. And I usually live in Linux. I have to constantly remind myself where I am when I am on Windows with all those almost-Linux consoles ...
20161010 05:05:59< wedge009> Ah, don't worry. More than once I've found myself typing dir when I meant ls and vice versa. x.x
20161010 05:06:52< Aginor> powershell supports ls
20161010 05:07:09< Aginor> in my opinion, that's the main advantage it has
20161010 05:07:10< tad_> My issue today was not realizing I'd forgotten to update me git repo and pushing up comits which had been deleted.
20161010 05:08:14< tad_> Powershell used to be a mess, I thought, so I've stayed away from it. Good ol' DOS Command shell serves well enought for the little I do on Windows.
20161010 05:08:58< Aginor> powershell is a modified version os zsh so it's kind of nicer
20161010 05:09:07< wedge009> I haven't need to look at Powershell yet, but might for if/when I ever need to work with an AWS Windows system.
20161010 05:09:50< Aginor> it seems a bit neat how it's linked into being able to run random c# things, but it also strikes me as a potential security issue
20161010 05:10:24< wedge009> Heh, oh dear.
20161010 05:11:06-!- Kwandulin [~Miranda@p200300760F2C71D3D56812C2C9812869.dip0.t-ipconnect.de] has quit [Read error: Connection reset by peer]
20161010 05:14:17< tad_> Powershell? That's right .. when it started it was 'Don't use, it's a major risk" and a lot of us disabled it because it was such a broad attack surface. That was long ago, so I presume it's been (mostly) fixed.
20161010 05:14:43-!- Nobun [~nobun@5.170.105.159] has joined #wesnoth-dev
20161010 05:15:07< Nobun> shaowm: are you active?
20161010 05:15:26< Nobun> shadowm *
20161010 05:15:37< tad_> Patience. He was active a while ago.
20161010 05:16:51< Nobun> tad_: I corrected myself becouse I miswrote the name. Writing correctly the name the irc client could notice he was called :P
20161010 05:17:09< Nobun> but thank for the information ;)
20161010 05:17:52 * tad_ nods. I spoke too quickly and didn't notice the typo.
20161010 05:18:49< Nobun> naaah... don't worry :) The intention was appreciable :)
20161010 05:19:54< tad_> Sheesh. Reading the CodeBlocks install page. Are Borland or Digital Mars compilers even still maintained and available?
20161010 05:22:03< Nobun> I think Borland compilers should be availabe, but I don't have any news wich can confirm it
20161010 05:22:12< Aginor> tad_: wikipedia says yes
20161010 05:22:32< Aginor> current incarnation: https://en.wikipedia.org/wiki/C%2B%2BBuilder
20161010 05:23:37< tad_> It's been ages since I've even considered them, or Intel. Probably around 1990.
20161010 05:24:35< Aginor> intel I'd consider more
20161010 05:24:40< tad_> I remember I had to set up a build farm to ensure a product could support all of 'em.
20161010 05:24:48< Aginor> they still kick butt in the high performance computing space
20161010 05:25:24< tad_> Oh, yeah. That's why I didn't ask about Intel .. I know it's still active and heavily used. But .. "Borland" ???
20161010 05:26:31< Aginor> Turbo C++ → Borland C++ → Borland C++Builder → CodeGear C++Builder → Embarcadero C++Builder
20161010 05:27:09< Aginor> mind you, I know of at least one company that is still actively developing in Delphi
20161010 05:28:35< Aginor> (from a rather small set of companies considered)
20161010 05:28:40< tad_> I gave up on Borland when they turned Pascal into a mess. But, then, I learned Pascal from the GTA who did most of the work on Pascal for Wirth.
20161010 05:28:45< Nobun> Aginor, tad_ : however, searching around the net, I am unable to find Borland official site
20161010 05:28:48< Aginor> so I guess the borland tools are still living
20161010 05:29:26< Aginor> Nobun: Borland went belly up, all their stuff got bought by CodeGear, who went bellyup and Embarcadero bought it all
20161010 05:29:30< tad_> That's what I tought .. Borland was bought like 1995 and their sites shut down around 2000, IIRC.
20161010 05:29:30< Nobun> Or, better, not an official site where c++ is mentioned (borland.com seems to show medical machines and not programming languages)
20161010 05:29:34< Aginor> or maybe CodeGear is still around
20161010 05:29:36 * Aginor shrugs
20161010 05:29:53< Nobun> Aginor: thank for the information
20161010 05:29:56< Aginor> Nobun: Borland, the software company, is gone since 2007ish
20161010 05:30:31< Nobun> I figure that happened when Visual C++ becomes free
20161010 05:30:51< Nobun> (wich happened around 2005 if I remember well)
20161010 05:30:56< Aginor> https://en.wikipedia.org/wiki/Borland
20161010 05:31:30< Aginor> Nobun: I don't think C++ was ever their biggest platform, I think Pascal and later Delphi were
20161010 05:31:39< Aginor> and they never became free
20161010 05:32:50< Nobun> personally I HATE delphi and pascal :P
20161010 05:33:50< Aginor> I cut my programming teeth on pascal :D
20161010 05:34:42< tad_> Pascal is OK if you ignore the fact the original language from Wirth could not even DO input or output. It's a "purist" language and, having studied and worked under his GTA I always felt it was not so much meant to be used but to teach .. but really more to make a statement that he didn't think much of the direction Algor and PL/1 went.
20161010 05:35:14< tad_> s/Algor/Algol/
20161010 05:35:54< Nobun> Well... I am not a programmer at all... this is probably the reason why I hate pascal. All I know about C++ / python programming is a self-learning process, only becouse I love programming since I was 13 years old
20161010 05:36:54< tad_> Well, you do C++ so that's sufficient (although not necessary) to prove you can't program :P
20161010 05:37:17< Nobun> lol
20161010 05:37:48< Aginor> I'm not a big fan of pascal either to be honest
20161010 05:38:01< Nobun> well... what I can say is... the fact I am not good on programming is I am unable to understand 90% of wesnoth source code XD
20161010 05:38:02< Aginor> clear syntax for learning, but that's about it
20161010 05:38:36< Aginor> Nobun: it could also be that it's reasonably big and complex and is big on side effects and tightly coupled behaviour
20161010 05:39:59< tad_> I wonder what Wirth would do if he started over and did Pascal based on C++, today.  "Sure, you can do classes, but NO side effects .. and forget about I/O or threads!"
20161010 05:40:17< Nobun> yeah, sure... but I had difficulties also when I tried to fix myself a specify bug wich was easy to find in source. Infact I found where I needed to apply changes, but I was unable to apply a 100% working solution on that code structure (so why I never added a PR for that bug)
20161010 05:40:22< shadowm> Nobun: Yes?
20161010 05:40:44< Nobun> shadowm: I have some question about the abandoned WesCamp project
20161010 05:40:51< Nobun> I need *
20161010 05:40:58< Nobun> sorry
20161010 05:41:24< Nobun> More in details, I need to understand how wescamp was able to upload translations at gettext.wesnoth.com
20161010 05:41:59< Nobun> I'm planning to make a python GUI utilities wich can help UMC developers to create and maintain translations, using wmlxgettext, msginit, msgmerge under the scene
20161010 05:43:37< shadowm> Nobun: I know that WesCamp did not _upload_ things, but rather a cronjob at gettext.wesnoth.org pulled from it.
20161010 05:44:44< shadowm> It involves the codebase here, starting with bin/update-gettext-stats: https://github.com/wesnoth/website/tree/master/gettext.wesnoth.org
20161010 05:45:15< shadowm> (Content warning: PHP that's so bad, even people who can tolerate PHP would pluck their eyeballs out.)
20161010 05:45:37-!- esr [~esr@static-71-162-243-5.phlapa.fios.verizon.net] has joined #wesnoth-dev
20161010 05:45:37-!- esr [~esr@static-71-162-243-5.phlapa.fios.verizon.net] has quit [Changing host]
20161010 05:45:37-!- esr [~esr@wesnoth/developer/esr] has joined #wesnoth-dev
20161010 05:46:44< Nobun> I never learnt PHP :P
20161010 05:47:13< shadowm> I advise you don't unless your income ever depends on it.
20161010 05:47:15< Nobun> sorry for the dumb question, but what is a cronjob?
20161010 05:47:42< shadowm> Unix jargon for automated tasks that can be scheduled to run periodically.
20161010 05:48:10< Nobun> yeah, but if you need to write down a dinamic site from scratch (or to modify PHP part of a CMS) probably PHP is still the best language to use.
20161010 05:48:25< shadowm> The daemon that's usually in charge of managing that is usually cron or a cron lookalike, hence cron.
20161010 05:48:31< Nobun> I don't know why, but I always find ASP even worst
20161010 05:48:40< Aginor> Nobun: I'd suggest python + flask or python+django nowadays
20161010 05:49:11< Nobun> Aginor: yeah you are right, but (at least in Italy) it is hard to find a provider wich allows to use python
20161010 05:49:15< shadowm> (Related: crontab is what the file (usually plain text) with the tasks schedule table is called.)
20161010 05:49:20< Aginor> they're all saner than php
20161010 05:49:30< Aginor> Nobun: hosting provider?
20161010 05:49:40< tad_> Nobun, you don't 'learn' PHP you just sorta hack at it until is seems to work. :P
20161010 05:49:40< Nobun> and a lot of famous CMS (like joomla and drupal) still uses PHP
20161010 05:50:23< Nobun> Aginor: I mean... if you plan to develop a site, you probably want it works when uploaded in internet
20161010 05:50:42< shadowm> <tad_> Nobun, you don't 'learn' PHP you just sorta hack at it until is seems to work. :P <-- And when you least expect it, you've given the whole world authorization to steal your private keys and credit card number.
20161010 05:51:00< Nobun> shadowm: yes I know
20161010 05:51:11< shadowm> Or if not that, at least turn yourself into an unwilling participant of the malware distribution machine.
20161010 05:51:40< Nobun> the main reason why I never tried to learn PHP is that I doubt I can create a secure code for vulnerable actions like authentications and similar
20161010 05:51:47 * tad_ waves.
20161010 05:51:50-!- tad_ [~lundberg@173.217.65.103] has quit [Quit: Leaving]
20161010 05:51:51< Aginor> Nobun: the biggest gripe with PHP is that the API is counter-intuitive and makes it easy to make insecure solutions, last time I looked, the official documentation itself led you to insecure implementations
20161010 05:52:00< Aginor> thus, I'm not a big supporter of PHP
20161010 05:52:15< Nobun> Yes, I agree with your doubt
20161010 05:52:16< shadowm> The API is essentially Wesnoth in spades.
20161010 05:52:47< Aginor> Nobun: if you want to look at something more sensible, I'd suggest django+python or flask+python, have a look at heroku as a hosting provider
20161010 05:52:51< Aginor> or something similar
20161010 05:53:07< Nobun> however, coming back to gettext.wesnoth.org, is there any protocols wich can be directly used on a python progam?
20161010 05:53:29< shadowm> Python's standard library comes with almost everything I think?
20161010 05:54:20< Nobun> probably. Python standard library, for example, has a library for FTP for example
20161010 05:54:37< Aginor> it wouldn't expose a rest service, would it?
20161010 05:55:27< Nobun> I don't know... in this moment the python GUI I'm thinking to develop isn't yet started
20161010 05:56:18< Nobun> the idea is to have a GUi wich do, more or less... create blank po, update and merge translations, upload merged translations to allow translator to edit and upldate pot files
20161010 05:57:07< Nobun> more or less the same tasks wich WesCamp wanted to do
20161010 05:57:28< shadowm> Huh...
20161010 05:57:46< shadowm> Translators aren't supposed to "edit" or "update" pot files, only po files.
20161010 05:58:06< shadowm> vultraz: Is there a way to have a listbox without an active selection?
20161010 05:58:15< Nobun> yes... probably I explained badly my point
20161010 05:58:59< Nobun> I mean... the developer of campaign could use the GUI to easily create and update pot files (wmlxgettext, msginit, msgmerge) without having to know how to use those tools directly
20161010 05:59:29< shadowm> Sounds good so far.
20161010 05:59:48< Nobun> the "updated" pot (updated with the changes of CAMPAIGN/addon) but without the new translations could be uploaded for translators, wich will be able later to actually update the translations
20161010 06:00:42< shadowm> That's when the pot is merged into a po (or a blank slate po is created from the pot).
20161010 06:02:35< Nobun> Yes... the idea is... perhaps the final translator will simply use PoEdit (or another software) to apply the translations. But we can make his work easier, if the UMC author can upload a pot already merged
20161010 06:02:47< Nobun> the process I am thinking about is
20161010 06:03:21< shadowm> Upload to where exactly?
20161010 06:04:07< Nobun> creating skele.po (blank pot) --> msginit po into different blank POT for every translations (when plural forms used) --> merge blank msginited pots with pastly translated pots (for every language) --> upload the final pots obtained this way
20161010 06:04:27< Nobun> the idea is to upload at gettext.wesnoth.org, if possible
20161010 06:05:45< Nobun> obliuouvsly this is NOT a project I am thinking to be unofficial... becouse, if I can develop it but it is unsercure for some reason, I'd like you can check it before approving it
20161010 06:05:53< shadowm> You mean blank po, not blank pot. pots are by definition blank (if blank means no translated strings).
20161010 06:06:13< Nobun> yeah, shadowm
20161010 06:06:35< vultraz> has_minimum = false
20161010 06:07:06< shadowm> The way things originally worked (wen they worked) were more or less as you say, except without anyone having to do anything by hand.
20161010 06:07:21< Nobun> exactly, shadowm 
20161010 06:07:31< Nobun> all those steps are not made by hand
20161010 06:07:51< shadowm> EXCEPT that add-on authors still had to compile the final pos and somehow discern which ones needed to be updated in their add-ons.
20161010 06:08:12< Nobun> it is running wmlxgettext (in a easier way than using from command line), using msginit (to setup nplurals correclty if plurals forms used) and using msgmerge (to merge translations)
20161010 06:08:33< shadowm> So um, I don't see what could possibly be insecure about anything you say. That's really a matter of the implementation.
20161010 06:08:48< shadowm> The problem is that you are still delegating an extra burden to add-on maintainers.
20161010 06:09:11< shadowm> Which is the part where they have to manually get their add-ons into some kind of interface rather than just mark translate=true in their .pbls.
20161010 06:10:24< shadowm> vultraz: How would I go about making it so the code can deselect all entries, but the user *cannot*?
20161010 06:10:34< Nobun> I understood. My objective is to make life simplier to all, so I am open to review or abandone my idea
20161010 06:11:20< vultraz> shadowm: what is the usecase?
20161010 06:12:03< vultraz> either way, I don't think it's possible
20161010 06:12:06< shadowm> Nobun: So I don't remember exactly why WesCamp ended up in this state where someone (AI0867) had to somehow crank it by hand in order to update the add-on files from the add-ons server into the WesCamp repositories and run the whole wmlxgettext show afterwards.
20161010 06:12:09< vultraz> without manual intervention
20161010 06:12:47< shadowm> Nobun: But one aspect to consider is that wmlxgettext (at least the original version) can't deal with some border cases, and GNU gettext's tools (msgfmt etc.) can't either.
20161010 06:13:08< shadowm> So when you automate that, things are likely to fail for any of the following reasons, if not more:
20161010 06:13:47< shadowm> 1) Author used an illegal textdomain (e.g. WesCamp mandated "wesnoth-Addon_Directory_Name" and some people insisted on using "wesnoth-ADDONACRONYM" like in mainline)
20161010 06:14:02< shadowm> 2) Author didn't declare a textdomain for their add-on
20161010 06:14:11< shadowm> 3) Author didn't use any translatable strings
20161010 06:14:12< Nobun> shadowm: currently wmlxgettext try to be as accurate as possible, but obliouvsly I cannot guarantee it is perfect
20161010 06:14:21< shadowm> 4) Author didn't bind any translatable strings to their add-on's textdomain
20161010 06:14:50< shadowm> 5) Author has an empty translatable string in their code (the empty msgid is used by GNU gettext for bookkeeping purposes)
20161010 06:15:18< shadowm> 6) Author used an encoding that is incompatible with our translation toolchain (read: everything that isn't BOMless UTF-8)
20161010 06:15:34< Nobun> case 5) is solved by wmlxgettext current version IGNORING all empty translatable string (not anymore closing translation becouse the error)
20161010 06:16:12< shadowm> That's what the original wmlxgettext should've done in the first place, and additionally I've always thought the game ought to issue warnings to stderr as well.
20161010 06:16:14< Nobun> you can notice wmlxgettext currently returns a warning on a wesnoth core file, wich uses an empty translatable string _" "
20161010 06:16:47< shadowm> I haven't run the master version of wmlxgettext since I'm not in charge of the release process for that branch anymore.
20161010 06:17:32< shadowm> I also think that the textdomain naming scheme requirements are 1) not properly documented; 2) not enforced by the game; 3) pedantic as hell.
20161010 06:18:49< Nobun> case 6) is not solved directly by wmlxgettext (and can't)
20161010 06:18:54< shadowm> I believe the "wesnoth-" prefix in the first place only exists so that our translations can coexist with other program's in <INSTALL PREFIX>/share/translations or so.
20161010 06:19:08< shadowm> But that obviously doesn't apply to add-ons, and really needn't apply to mainline either.
20161010 06:19:50-!- vultraz_iOS [uid24821@wesnoth/developer/vultraz] has quit [Quit: Connection closed for inactivity]
20161010 06:20:07< shadowm> Making the WML parser implicitly add the prefix if it's so desirable would've been a better choice than imposing a seemingly arbitrary requirement on people who may probably not even know how to write a factorial function.
20161010 06:20:36-!- JyrkiVesterinen [~JyrkiVest@nblzone-242-23.nblnetworks.fi] has joined #wesnoth-dev
20161010 06:21:50< shadowm> But this goes hand-in-hand with my point about the PHP API. Wesnoth was put together by lots of different people with different ideas and most of them without any concrete long-term goals, ergo no maintainability standards.
20161010 06:22:35< shadowm> And the core of the v1.11.18 incarnation of the i18n machinery dates as far back as 2004.
20161010 06:22:48< shadowm> Late 2004/early 2005, at least.
20161010 06:23:06< shadowm> </rant>
20161010 06:24:17< shadowm> I think (again, not sure, AI0867 did this stuff, I only watched from afar) two other problems with the latest incarnation of WesCamp was the storage model.
20161010 06:24:59< shadowm> For one, it created one repository for each add-on/server instance pair on GitHub: https://github.com/wescamp
20161010 06:25:40< shadowm> I believe this, coupled with the half-hourly schedule of the gettext.wesnoth.org generation cronjob, caused the script to run into GitHub API limits a few times.
20161010 06:26:29< shadowm> Secondly, IIRC some manual preparation was still needed for _new_ add-ons published with translate=yes on the server that weren't previously on WesCamp.
20161010 06:27:24< shadowm> And then AI0867 drifted away and essentially disappeared without leaving anyone in charge of this mess.
20161010 06:28:14< shadowm> I eventually acquired admin rights on WesCamp at some point around Q3 last year, but guess what, I too drifted away before I could even begin to figure out what the actual problem that brought WesCamp to a halt circa 2014 was.
20161010 06:29:05< shadowm> (I'm open to promoting new admins there, for any adventurous souls who feel they can somehow figure out how to make it work again without any documentation or guidance.)
20161010 06:30:31< shadowm> So if you wanted to avoid replicating WesCamp's disaster you'd need to come up with a solid automation scheme that doesn't require any active work from an admin. And documentation.
20161010 06:31:34< shadowm> You'd probably also want to make it so WesCamp itself doesn't need to hold a whole copy of each add-on there, only the translations. Or at least blacklist the pointless garbage (.ogg, .wav, .png, .jpg, and pretty much anything that doesn't look like a UTF-8 text file).
20161010 06:31:54< shadowm> s/WesCamp/WesCamp's successor/
20161010 06:32:59< shadowm> If it all were to be entirely hosted on wesnoth.org's facilities, this would be pretty much mandatory, since pointless duplication of data files is a big no-no on my watch.
20161010 06:33:38< shadowm> Not because we can't afford the disk space, but rather because we can't afford the performance cost every time some dumb tool needs to somehow look at the files' contents.
20161010 06:34:40< shadowm> After all, all of this crap runs on the same machine and demands time from the same RAID1 array.
20161010 06:35:36< shadowm> Website everything, database, campaignd (add-ons server) instances, wesnothd (MP server) instances under the server.wesnoth.org hostname, Doxygen generation, gettext.wesnoth.org generation, units.wesnoth.org generation.
20161010 06:35:50< shadowm> Most of them pretty disk-intensive tasks.
20161010 06:36:47< Nobun> hmm I understood
20161010 06:37:50< Nobun> yeah... the reason why I asked about 'upload' is becouse the idea was to find a way to don't require an admin action to make use of upload translations
20161010 06:38:33< Nobun> but, obliouvsly, even if this task could be performed, it may open other downside problems too 
20161010 06:39:57< shadowm> Everything has at least one downside to it (see also: living kills people).
20161010 06:46:33< shadowm> What on earth is scons smoking? http://pastebin.com/ueSP2GXj
20161010 06:47:07< shadowm> I just added actual code to that file, and it's part of the wesnoth target. How is wesnoth "up to date"???
20161010 06:48:04< shadowm> Oh.
20161010 06:48:51< shadowm> I guess scons noticed that the compiler's output was identical due to optimizations.
20161010 06:49:14< shadowm> Namely, the optimization that doesn't require it to emit code for if(foo & 0) { /* this is never reached */ }
20161010 06:50:51< shadowm> vultraz: https://dl.dropboxusercontent.com/u/21371130/screenshots/Screenshot_20161010_035023.png 
20161010 06:51:19< shadowm> vultraz: I want the user to be able to click on the bookmarks on the left to select one. Naturally, the user shouldn't be able to select them because that's not a meaningful action.
20161010 06:51:33< shadowm> ... not be able to deselect them.
20161010 06:52:00< shadowm> vultraz: The code should be still able to deselect them, however, to signify that the user browsed away from a bookmark.
20161010 06:53:27< shadowm> Ideally I'd accomplish this without having to tamper with event handling more than I already did for the file list, which already took like 45% of the dialog's LoC count.
20161010 06:53:33< vultraz> shadowm: ok, I think you want has_minimum = false and a hack such as this: https://github.com/wesnoth/wesnoth/blob/master/src/gui/dialogs/drop_down_list.cpp#L37
20161010 06:54:48< shadowm> Deer god.
20161010 06:55:12< shadowm> Why does the answer always have to be more custom event handling.
20161010 06:56:02< shadowm> No, don't answer that rhetorical question. It has a stop instead of a question mark at the end for a reason.
20161010 06:56:25-!- atarocch [~atarocch@93.56.160.28] has joined #wesnoth-dev
20161010 07:00:21< shadowm> vultraz: Completely unrelated to that, do you think that the .. entry's purpose is easily understood/deduced, or should it use an icon and a new label?
20161010 07:00:39< shadowm> And also, should files have icons too?
20161010 07:00:57-!- Nobun [~nobun@5.170.105.159] has quit [Quit: Salve a tutti]
20161010 07:01:08< shadowm> If the answer to either is yes we'll need icons more in line with LordBob's style, also for directories.
20161010 07:01:31-!- atarocch [~atarocch@93.56.160.28] has quit [Quit: Leaving]
20161010 07:02:29< shadowm> (And no, I don't want to discuss the aesthetics of the places list on the left until I'm finished with the code. It should go without saying that this isn't anywhere near its final form.)
20161010 07:02:45< vultraz> well, we at least have an immediate replacement for the folder icon
20161010 07:02:48< shadowm> (Consider it programmer art for now.)
20161010 07:03:06< shadowm> Do we?
20161010 07:04:14< shadowm> I don't see anything in the icons dir that looks remotely like a folder (which is why the browse buttons on the Version dialog and the screenshot notification use those tacky arrows).
20161010 07:04:46< shadowm> Wait a second.
20161010 07:04:51< vultraz> the Load icon?
20161010 07:04:52< vultraz> from the editor?
20161010 07:04:55< vultraz> sans the arrow
20161010 07:05:26< shadowm> mind.state() == BLOWN
20161010 07:05:37< shadowm> No but seriously why didn't I think of this years ago?
20161010 07:05:43< shadowm> YEARS.
20161010 07:06:11< vultraz> what?
20161010 07:06:23< shadowm> The Load icon.
20161010 07:06:25< shadowm> From the editor.
20161010 07:06:27< shadowm> vultraz.
20161010 07:06:36< vultraz> I just said that
20161010 07:06:37< shadowm> This has been around for much longer than those dialogs.
20161010 07:06:46< shadowm> Why didn't I think of editing them back then?
20161010 07:06:51< vultraz> I have no idea.
20161010 07:06:56< shadowm> Why didn't *you* suggest it to me? :\
20161010 07:07:05< vultraz> Because I didn't think of it either xD
20161010 07:07:10< vultraz> But I've just edited out the arrow
20161010 07:07:35< shadowm> I even told LordBob "yeah no we don't have any icons for what I need so I'm going to use the big arrow for now".
20161010 07:07:57< vultraz> the big...arrow... for a folder icon?
20161010 07:07:59< vultraz> what?
20161010 07:08:06< shadowm> FOR THE BROWSE BUTTONS.
20161010 07:08:39< vultraz> Should I commit this folder icon.
20161010 07:08:43< shadowm> Browse buttons ought to use folders, not the tacky arrows. That's what I very clearly implied above.
20161010 07:08:53< shadowm> Commit "this" folder icon? Do you mean you already have an edit done?
20161010 07:09:01-!- atarocch [~atarocch@93.56.160.28] has joined #wesnoth-dev
20161010 07:09:45< vultraz> https://drive.google.com/file/d/0B-mR9s8FduLLYXdpNlZWYVNUUWM/view?usp=sharing
20161010 07:10:03-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has joined #wesnoth-dev
20161010 07:10:06< shadowm> Yes, please.
20161010 07:11:03< irker158> wesnoth: Charles Dang wesnoth:master 090a2b7b1ef7 / images/misc/folder-icon.png: Updated folder icon https://github.com/wesnoth/wesnoth/commit/090a2b7b1ef777245f9b57f3f5abf64a314e5156
20161010 07:11:19< shadowm> Okay, now I need button versions of that icon.
20161010 07:11:37< shadowm> And wire them into...
20161010 07:12:31< shadowm> Into a clone of the action_go 25x25 button widget variant named action_browse.
20161010 07:12:45< shadowm> (No, you can't replace action_go because it's used for web stuff too.)
20161010 07:13:23< shadowm> If it's not too much trouble, that is. I could make them myself.
20161010 07:13:44< shadowm> I still know how to operate an image editor.
20161010 07:14:01< vultraz> Done
20161010 07:14:07 * vultraz adds wml
20161010 07:19:31-!- boucman_work [~boucman@gre92-5-82-237-199-7.fbx.proxad.net] has joined #wesnoth-dev
20161010 07:20:39-!- boucman_work [~boucman@gre92-5-82-237-199-7.fbx.proxad.net] has quit [Remote host closed the connection]
20161010 07:20:59-!- boucman_work [~boucman@fw-alt.idf.smile.fr] has joined #wesnoth-dev
20161010 07:24:09< irker158> wesnoth: Charles Dang wesnoth:master 25eb58abb698 / / (4 files in 2 dirs): New Browse button definition and images https://github.com/wesnoth/wesnoth/commit/25eb58abb69857c9a55dbdfeaa41aa5cc0370441
20161010 07:24:34< vultraz> shadowm: ^
20161010 07:25:08< vultraz> the folder icon is darker than the base button image, for the record.
20161010 07:26:49< shadowm> What?
20161010 07:27:14< vultraz> The folder icon is darker than icons/action/browse_25.png
20161010 07:27:23< vultraz> Just stating for the record.
20161010 07:27:39< shadowm> Ah. Well, if it makes sense, sure.
20161010 07:28:09< shadowm> Thanks for doing this. Won't be able to take a look right now though, too busy with code to stash and checkout and pull.
20161010 07:28:18< vultraz> Np
20161010 07:36:07< shadowm> Yessssssss.
20161010 07:36:47< shadowm> I got the bookmarks bar working, just missing the part where the user's actions on the fileview have an effect on the bookmark selection.
20161010 07:37:40< shadowm> Didn't even need that much code to set it up.
20161010 07:56:15< irker158> wesnoth: Charles Dang wesnoth:master 1f1855300dad / / (15 files in 5 dirs): GUI2/Stacked Widget: removed [stack] wrapper around [layer] and use the latter d https://github.com/wesnoth/wesnoth/commit/1f1855300dad676b44f9b85aaed3f44a60954d92
20161010 07:56:58< shadowm> And I got the code done.
20161010 07:57:27< vultraz> tldr now we use [stacked_widget][layer] instead of [stacked_widget][stack][layer]
20161010 07:57:38< shadowm> This is going to be the greatest thing ever when it's done.
20161010 07:58:20< celticminstrel> shadowm: I think an icon for the ".." would be good, but I also think most people probably know what ".." means, so it's probably not necessary.
20161010 07:58:37< vultraz> it's rather intuitive 
20161010 07:58:56< shadowm> So the thing boils down to whether files should have icons too.
20161010 07:59:02< celticminstrel> What's intuitive?
20161010 07:59:38< celticminstrel> vultraz: Did you see my code comment?
20161010 07:59:48< irker158> wesnoth: Charles Dang wesnoth:master 9fa82c03d499 / src/gui/widgets/stacked_widget.cpp: Fixup doc comment from 1f1855300dad https://github.com/wesnoth/wesnoth/commit/9fa82c03d499bf0bcf5da1cffd5ee6d8df49c856
20161010 08:00:52 * celticminstrel spots gfgtdf's RELEASE_NOTES update and remembers that someone needs to enable scenario_generator=lua
20161010 08:02:27< irker158> wesnoth: Charles Dang wesnoth:master 4cc22e125892 / src/help/help_impl.hpp: translation::compare takes string arguments, no need to call it with .c_str() https://github.com/wesnoth/wesnoth/commit/4cc22e1258920b7003fbabd7e1752a0d7373c794
20161010 08:02:32< vultraz> celticminstrel: that one?
20161010 08:02:38< celticminstrel> Yes.
20161010 08:02:53< shadowm> These icons are simple enough that I could whip up a generic-sheet-of-paper icon if you want.
20161010 08:03:37 * shadowm considers marking symbolic links in italics and highlighting specials in red.
20161010 08:03:53< celticminstrel> I'm kind of against using a generic file icon.
20161010 08:04:17< shadowm> So what icon would you want instadead?
20161010 08:04:24< celticminstrel> Ideally file icons should reflect the file type, but that's probably too much work for little gain here.
20161010 08:04:37< celticminstrel> I don't really mind them not having an icon.
20161010 08:04:38< shadowm> It's a colossal amount of work, yes.
20161010 08:04:46< celticminstrel> What do you mean by specials?
20161010 08:05:14< shadowm> Named pipes, sockets, character devices, block devices.
20161010 08:05:19< celticminstrel> Speaking of symbolic links, it'd be great to equivalently handle aliases and shortcuts. (No idea if it's worthwhile.)
20161010 08:05:25< shadowm> Or anything that isn't a directory or a regular file or a symbolic link.
20161010 08:05:31< shadowm> What the hell is an alias?
20161010 08:05:49< shadowm> No, I'm not interested in Windows-specific crap like shortcuts.
20161010 08:05:51< celticminstrel> It's a Mac thing.
20161010 08:05:58< shadowm> Neither, then.
20161010 08:06:12< celticminstrel> Actually, it'd be best if Boost does this internally...
20161010 08:06:24< shadowm> I don't see why it would.
20161010 08:06:45< shadowm> Windows shortcuts are a graphical shell construct and Windows' core API itself doesn't treat them specially.
20161010 08:07:17< celticminstrel> I think the same is true of Mac aliases, but not quite sure. Certainly the command-line doesn't treat them specially.
20161010 08:07:43< shadowm> Whereas symbolic links and junction points are ingrained within the filesystem.
20161010 08:09:00< shadowm> Reminds me I want to make the Delete option call a function that doesn't stupidly dereference symbolic links and destroy everything inside.
20161010 08:09:22< shadowm> I lost my campaign's files once thanks to a similar issue with the add-ons manager code in 1.3.x.
20161010 08:12:15 * shadowm hasn't failed to realize that the current UI vigilante keeps ignoring my inquiry wrt file icons.
20161010 08:12:57< vultraz> I don't think it's necessary
20161010 08:14:31< shadowm> Okay then.
20161010 08:15:33< vultraz> Since we cannot have individual icons for file types, it's better to have no icon that a 'text' icon for everything.
20161010 08:15:55< shadowm> The generic sheet of paper has been in use as a generic file icon for decades, not just for text.
20161010 08:16:02< shadowm> Kids these days.
20161010 08:16:07< celticminstrel> Yeah, the generic file icon is not a "text" icon.
20161010 08:16:17< vultraz> Still, it might crowd the aesthetic.
20161010 08:16:19< celticminstrel> A "text" icon is a generic file icon plus lines of text.
20161010 08:16:41< celticminstrel> Well, stuff that looks like it might have been text that was shrunk to illegibility.
20161010 08:17:05-!- celticminstrel [~celmin@unaffiliated/celticminstrel] has quit [Quit: And lo! The computer falls into a deep sleep, to awake again some other day!]
20161010 08:17:40< vultraz> Let's see what AtS looks like in my new Mp Staging dialog
20161010 08:20:14< shadowm> https://dl.dropboxusercontent.com/u/21371130/screenshots/Screenshot_20161010_051934.png
20161010 08:38:54< vultraz> oh my god that UI
20161010 08:39:09< vultraz> THE PAIN
20161010 08:39:16< vultraz> (but i see what you're saying, yes)
20161010 08:40:44< vultraz> ok, now to re-add the Entry Points feature
20161010 08:41:43< vultraz> where should I add it...
20161010 08:41:48< vultraz> pre-Staging, I guess..
20161010 08:43:27< vultraz> I don't really want to construct a configure_engine, though...
20161010 08:43:55< vultraz> let's pluck the functionality 
20161010 08:45:43-!- Ivanovic_ [~ivanovic@p579FBF3F.dip0.t-ipconnect.de] has joined #wesnoth-dev
20161010 08:46:47-!- JyrkiVesterinen [~JyrkiVest@nblzone-242-23.nblnetworks.fi] has quit [Quit: .]
20161010 08:49:39< Rhonda> Anyone going to be at 33c3?
20161010 08:49:44-!- enchi_ [enchilado@defocus/yummy/enchilado] has joined #wesnoth-dev
20161010 08:49:55-!- Aginor_ [~andreas@apollo.alternating.net] has joined #wesnoth-dev
20161010 08:50:16-!- timotei_ [~timotei@wesnoth/developer/timotei] has joined #wesnoth-dev
20161010 08:50:21-!- Soliton [~Soliton@wesnoth/developer/soliton] has quit [Disconnected by services]
20161010 08:50:27-!- Soliton_ [~Soliton@wesnoth/developer/soliton] has joined #wesnoth-dev
20161010 08:50:54-!- Soliton_ is now known as Soliton
20161010 08:54:51-!- Netsplit *.net <-> *.split quits: timotei, enchi, APic, aidanhs, Aginor, Appleman1234, irker158, Ivanovic
20161010 08:55:37-!- Ivanovic_ is now known as Ivanovic
20161010 09:01:20-!- Netsplit over, joins: APic
20161010 09:01:50-!- Netsplit over, joins: aidanhs
20161010 09:10:03< shadowm> vultraz: It still isn't possible to grow listboxes beyond their minimum size, right?
20161010 09:10:16< shadowm> So that the background can occupy the whole cell, that is.
20161010 09:11:34< shadowm> Well, I guess the listboxes do grow to fill up the cell, per se. I see their scrollbars do.
20161010 09:11:39< shadowm> Just not their backgrounds. :\
20161010 09:13:16< shadowm> https://dl.dropboxusercontent.com/u/21371130/screenshots/Screenshot_20161010_061259.png
20161010 09:13:44< shadowm> (Yes, I'm still moving everything around.)
20161010 09:13:54-!- Appleman1234 [~Appleman1@KD106154000208.au-net.ne.jp] has joined #wesnoth-dev
20161010 09:20:02-!- JyrkiVesterinen [~JyrkiVest@nblzone-242-23.nblnetworks.fi] has joined #wesnoth-dev
20161010 09:20:22< shadowm> And I broke the entire layout now.
20161010 09:20:56-!- Duthlet [~Duthlet@dslb-188-104-253-155.188.104.pools.vodafone-ip.de] has joined #wesnoth-dev
20161010 10:01:38< vultraz> shadowm: what?
20161010 10:02:58< vultraz> shadowm: the backgrounds are part of the toggle panels
20161010 10:03:02< vultraz> not the listbox
20161010 10:05:10< vultraz> shadowm: I actually investigated this before, and I recall at least 1 new definition each would need to be added to toggle panels and listboxes
20161010 10:06:25< vultraz> or perhaps just the listbox..
20161010 10:07:19< shadowm> Okay, the second line answers all.
20161010 10:07:31< shadowm> https://dl.dropboxusercontent.com/u/21371130/screenshots/Screenshot_20161010_070637.png
20161010 10:08:14< shadowm> I might or might not add the ability to add custom bookmarks.
20161010 10:08:37-!- bumbadadabum [~bumbadada@wesnoth/developer/bumbadadabum] has joined #wesnoth-dev
20161010 10:09:18< vultraz> shadowm: might recommend a 10 height space above the text box, actually
20161010 10:09:27< vultraz> besides that, it looks *awesome*
20161010 10:10:08< shadowm> For a few seconds the Delete button was an icon button.
20161010 10:10:26< vultraz> using the trash definiton?
20161010 10:10:39< shadowm> Uh, yeah, the trash definition that doesn't exist.
20161010 10:10:55< shadowm> But I did use the trash icon for that definition, yes.
20161010 10:11:05< vultraz> ...I thought I added that..
20161010 10:11:15< vultraz> must've never made it in
20161010 10:11:23< shadowm> So I decided against it for two reasons: the button is perhaps a little too small? Undecided as to whether this poses a problem or not.
20161010 10:11:27< vultraz> I was planning on inlining a delete button in Game Load
20161010 10:11:35< shadowm> The other reason is that the trash might be misleading.
20161010 10:11:40< vultraz> couldn't figure out event handling at the time
20161010 10:12:14< shadowm> We just call plain FS delete functionality, not whatever the platform's equivalent of the Trash/Recycle Bin is.
20161010 10:12:50< vultraz> is there even a way to do that
20161010 10:13:37< shadowm> Yes, on Windows
20161010 10:14:45< shadowm> I don't know anything about OS X, and I don't really know how it works on X11 but I believe you are just supposed to move your crap to the Trash directory if it exists.
20161010 10:14:51< JyrkiVesterinen> On GNU/Linux there is a program called trash-cli. IDK how easy/hard it would be to replicate its functionality.
20161010 10:14:52< JyrkiVesterinen> https://github.com/andreafrancia/trash-cli
20161010 10:15:04< shadowm> i.e. on X11 there's no API.
20161010 10:15:48< shadowm> Okay, actually, there is an "API".
20161010 10:16:05< vultraz> hmm
20161010 10:16:15< shadowm> fdo has a Trash spec.
20161010 10:16:18< vultraz> I forgot i had a configure engine member in Create..
20161010 10:16:26< shadowm> It's actually more elaborate tha I expected.
20161010 10:16:36-!- enchi_ is now known as enchi
20161010 10:16:42< shadowm> Goes to show that I almost never use the Trash functionality.
20161010 10:17:07< shadowm> (Not even when I was a Windows user, before 2005.)
20161010 10:18:17< shadowm> I'd rather come up with an icon that conveys the current implementation more accurately than waste months trying to make sense of how each platform approaches the Trash concept.
20161010 10:19:19< vultraz> I shall ask lordbob
20161010 10:19:39< vultraz> Convenient having a chat with him, it is.
20161010 10:19:40< shadowm> So, is there any value in adding a Rename option?
20161010 10:20:06< shadowm> The Yodaspeak thing is still so old it's probably decayed into a lighter element.
20161010 10:21:14< vultraz> (for the record, I have a private chat open with him on the Argentum Age room on Slack, which is how I quickly and easily speak to him without going through forum PMs)
20161010 10:21:30< vultraz> in Slack*
20161010 10:21:31< shadowm> (#ad)
20161010 10:21:50< vultraz> wait, no, it's on
20161010 10:21:56 * vultraz kicks self
20161010 10:22:23< shadowm> And independent of that, any other options...?
20161010 10:22:52< shadowm> I guess not.
20161010 10:23:27< shadowm> I could eventually add a file history pseudo-bookmark a la GTK.
20161010 10:23:45< shadowm> Or maybe not, actually, that'd complicate the rest of the workflow way too much.
20161010 10:24:02< vultraz> I dunno if rename is really necessary
20161010 10:24:07< shadowm> Almost all of the dialog assumes that it's dealing with paths that make sense.
20161010 10:24:07< vultraz> I think for now, not really.
20161010 10:26:00< vultraz> will you be committing soon?
20161010 10:26:08< shadowm> /home/shadowm/src/wesnoth/src/desktop/paths.cpp:73:81: error: cannot convert ‘WCHAR* {aka wchar_t*}’ to ‘LPCSTR {aka const char*}’ for argument ‘1’ to ‘WINBOOL GetVolumeInformationA(LPCSTR, LPSTR, DWORD, LPDWORD, LPDWORD, LPDWORD, LPSTR, DWORD)’
20161010 10:26:18< shadowm> Why on earth don't we just define UNICODE everywhere?
20161010 10:26:25< shadowm> It's 2016.
20161010 10:27:06< shadowm> I doubt there's any code that will silently break if ANSI calls and types become wide.
20161010 10:27:56< vultraz> blagh, my code crashes.
20161010 10:28:12< shadowm> vultraz: Is there a reason for asking that?
20161010 10:28:19< vultraz> no
20161010 10:28:39< vultraz> I'm just curious if this is forthcoming or still very much WIP
20161010 10:29:11< shadowm> I'm just missing the usual Windows test run.
20161010 10:29:53< shadowm> And organizing things into commits.
20161010 10:30:52< shadowm> And then I need to add the OS X equivalent of a function.
20161010 10:31:32< shadowm> So, 1 or 2 more days probably.
20161010 10:32:01< shadowm> Barring unexpected interruptions.
20161010 10:32:12< vultraz> ok, sweet
20161010 10:32:17< vultraz> will be in the release, then
20161010 10:35:00< vultraz> well
20161010 10:35:03< vultraz> this is odd p_p
20161010 10:36:44< vultraz> well, there's no way to get the scenario list in the Create dialog
20161010 10:36:56< vultraz> so..
20161010 10:36:57< vultraz> hm
20161010 10:37:54< shadowm>   debug desktop: enumerate_win32_drives(): E: is reported to be present
20161010 10:38:05< shadowm> Oh, never mind. I'm dumb.
20161010 10:38:18< vultraz> something's wrong with the sides..
20161010 10:38:24< shadowm> It shouldn't abort the enumeration when one drive seems dodgy.
20161010 10:39:11< shadowm> https://dl.dropboxusercontent.com/u/21371130/screenshots/Screenshot_20161010_073857.png
20161010 10:39:29< shadowm> As always, ignore the fact that ttext (or Cairo) doesn't work correctly on Wine.
20161010 10:40:00< vultraz> ahhh horizontal scrollbar!
20161010 10:40:02< shadowm> I need to decide how to deal with overlong bookmark labels. That one is a drive's label.
20161010 10:40:24< shadowm> Because yep, I went the extra mile and asked Windows for names.
20161010 10:41:38< vultraz> well, disable the horizontal scrollbar on that list
20161010 10:41:43< vultraz> for one
20161010 10:42:04< vultraz> I really do wish we had the ability to draw certain panels over other ones
20161010 10:42:44< vultraz> might be something to look into for 1.13.7
20161010 10:42:50< shadowm> vultraz: I think I shouldn't do that.
20161010 10:43:13< vultraz> then again, that would mostly only look good with sliding animations..
20161010 10:43:14< shadowm> https://dl.dropboxusercontent.com/u/21371130/screenshots/Screenshot_20161010_074303.png
20161010 10:43:17< vultraz> *le sigh*
20161010 10:43:40< vultraz> I'm going back to my code before I start another Wesnoth's APIs Suck rant.
20161010 10:44:50< shadowm> s/APIs/codebase/
20161010 10:45:18< shadowm> I'll just leave the horizontal scrollbar there.
20161010 10:45:38< vultraz> ok, I think I see why this is crashing...
20161010 10:45:38< shadowm> Also, there's no point in dropping the drive labels for this since the same situation will occur with translations of the hardcoded entries.
20161010 10:45:45 * vultraz curses
20161010 10:46:25< vultraz> essentially, the code in Create's post show
20161010 10:46:38< shadowm> Anyway, that's all for now.
20161010 10:46:41< vultraz> needs to be called after swapping the scenarios or something
20161010 10:47:15< shadowm> I just tossed the OS X "equivalent" of the "function" in there so I guess I'm pretty much done.
20161010 10:47:16< vultraz> but I cannot get the scenario list while the dialog is open
20161010 10:47:34< shadowm> Although I'd probably PR it since I'll want an OS X user to check something first.
20161010 10:47:36< vultraz> so maybe i need to move it to a function
20161010 10:47:37< shadowm> *'ll
20161010 10:47:50< vultraz> but what exactly calls the loadscreen
20161010 10:48:57< vultraz> I guess I'll have to ask gfgtdf some questions
20161010 10:50:00< vultraz> or..
20161010 10:50:02< vultraz> wait
20161010 10:50:18< vultraz> so the configure engine owns the status variables..
20161010 10:50:21< vultraz> hmm
20161010 11:03:17< zookeeper> shadowm, did you consider bookmark functionality, instead of having a lot of stuff in the places list by default (such as drives)?
20161010 11:08:21-!- enchi [enchilado@defocus/yummy/enchilado] has quit [Ping timeout: 268 seconds]
20161010 11:15:41-!- enchi [enchilado@defocus/yummy/enchilado] has joined #wesnoth-dev
20161010 11:25:57-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev
20161010 11:32:40-!- boucman_work [~boucman@fw-alt.idf.smile.fr] has quit [Ping timeout: 260 seconds]
20161010 11:39:45-!- Kwandulin [~Miranda@p200300760F2C7167B81AB181697B28FC.dip0.t-ipconnect.de] has joined #wesnoth-dev
20161010 12:32:45-!- Kwandulin [~Miranda@p200300760F2C7167B81AB181697B28FC.dip0.t-ipconnect.de] has quit [Read error: Connection reset by peer]
20161010 12:52:41-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has quit [Read error: Connection reset by peer]
20161010 12:53:12-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has joined #wesnoth-dev
20161010 13:03:13< DeFender1031> hmm... it looks like in 1.12, after victory, menu items added with [set_menu_item] will still show up in the menu, but not actually do anything.
20161010 13:07:52-!- Appleman1234_ [~Appleman1@KD106154010119.au-net.ne.jp] has joined #wesnoth-dev
20161010 13:09:11< vultraz> blah. where is gfgtdf when you need him
20161010 13:10:41-!- Appleman1234 [~Appleman1@KD106154000208.au-net.ne.jp] has quit [Ping timeout: 260 seconds]
20161010 13:13:30< DeFender1031> Let me test further though, it may be more complicated than that. (The item I noticed it with is one that brings us a [message] with [option]s. Want to see if some other action actually WOULD run.)
20161010 13:21:04-!- ancestral [~ancestral@75-168-189-115.mpls.qwest.net] has joined #wesnoth-dev
20161010 13:21:14-!- ancestral [~ancestral@75-168-189-115.mpls.qwest.net] has quit [Client Quit]
20161010 13:25:19-!- Appleman1234_ is now known as Appleman1234
20161010 13:39:08< vultraz> gfgtdf: so, if you come back, I was trying to add entry point selection back to the mp create game flow. i thought a simple item selector dialog would work, so I copied the fetch code from the configure_engine ctor and put it in enter_configure_mode, and it works, the scenario list comes up, but if you select anything other than the first scenario, the game crashes when launching the MP...
20161010 13:39:09< vultraz> ...Staging dialog. obv i need to update some setting to sync the gamestate but im not sure what
20161010 13:39:12< vultraz> gfgtdf: commit https://github.com/Vultraz/wesnoth/commit/a5e0688525233f989576a30634024d4b7bb0ad96
20161010 14:02:15-!- JyrkiVesterinen [~JyrkiVest@nblzone-242-23.nblnetworks.fi] has quit [Quit: .]
20161010 14:35:24-!- boucman_work [~boucman@gre92-5-82-237-199-7.fbx.proxad.net] has joined #wesnoth-dev
20161010 14:43:52-!- irker031 [~irker@uruz.ai0867.net] has joined #wesnoth-dev
20161010 14:43:53< irker031> wesnoth: mattsc wesnoth:master c91a65dde4c2 / data/ai/lua/ai_helper.lua: ai_helper check unit functions: also check if unit exists https://github.com/wesnoth/wesnoth/commit/c91a65dde4c21fccde7d417336c05ce6af590977
20161010 15:02:43-!- horrowind [~Icedove@2a02:810a:8380:10a8:21b:fcff:fee3:c3ff] has joined #wesnoth-dev
20161010 15:03:40-!- stikonas_ [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev
20161010 15:06:37-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Ping timeout: 240 seconds]
20161010 15:14:56< EliDupree> Interesting… (1.12) As far as I can tell, I've got a [get_global_variable] with side=2 in a side turn event, followed by a UI dialog in a turn refresh event – and the host player is stalled waiting for the global variable while the current players looking in the dialogue. Is that supposed to happen?
20161010 15:15:14< EliDupree> *client player is
20161010 15:19:26-!- Kwandulin [~Miranda@p200300760F2C716759298D27268B3352.dip0.t-ipconnect.de] has joined #wesnoth-dev
20161010 15:23:53-!- travis-ci [~travis-ci@ec2-54-167-15-199.compute-1.amazonaws.com] has joined #wesnoth-dev
20161010 15:23:54< travis-ci> wesnoth/wesnoth#11420 (master - c91a65d : mattsc): The build has errored.
20161010 15:23:54< travis-ci> Build details : https://travis-ci.org/wesnoth/wesnoth/builds/166448576
20161010 15:23:54-!- travis-ci [~travis-ci@ec2-54-167-15-199.compute-1.amazonaws.com] has left #wesnoth-dev []
20161010 15:37:18-!- tad_ [~lundberg@173.217.65.103] has joined #wesnoth-dev
20161010 15:41:45< EliDupree> Nevermind, I was incorrect about the order my code was in
20161010 16:01:32-!- Greg-Boggs [~greg_bogg@173.240.241.83] has joined #wesnoth-dev
20161010 16:01:47-!- Greg-Boggs [~greg_bogg@173.240.241.83] has quit [Remote host closed the connection]
20161010 16:01:53-!- Greg-Bog_ [~greg_bogg@173.240.241.83] has joined #wesnoth-dev
20161010 16:03:07-!- boucman_work [~boucman@gre92-5-82-237-199-7.fbx.proxad.net] has quit [Remote host closed the connection]
20161010 16:11:40-!- JyrkiVesterinen [~jyrki@78-27-105-35.bb.dnainternet.fi] has joined #wesnoth-dev
20161010 16:30:18-!- wedge010 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has joined #wesnoth-dev
20161010 16:30:18-!- celticminstrel [~celmin@unaffiliated/celticminstrel] has joined #wesnoth-dev
20161010 16:34:05-!- wedge009 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has quit [Ping timeout: 250 seconds]
20161010 16:34:05-!- wedge010 is now known as wedge009
20161010 16:40:12-!- ChipmunkV [~vova@static-89-94-113-91.axione.abo.bbox.fr] has joined #wesnoth-dev
20161010 16:45:18-!- stikonas_ is now known as stikonas
20161010 16:56:38< tad_> celticminstrel, What do you know about our C++ throwing 'lua_jailbreak_exception' or an exception class derrived from it?
20161010 16:59:39-!- gfgtdf [~chatzilla@x4e36a1e7.dyn.telefonica.de] has joined #wesnoth-dev
20161010 17:00:06< tad_> gfgtdf, What do you know about our C++ throwing 'lua_jailbreak_exception' or an exception class derived from it?
20161010 17:00:48< gfgtdf> tad_: well assume this call chain: c++ code -> lua code -> c++ code 
20161010 17:01:08< tad_> OK.
20161010 17:01:09< gfgtdf> tad_: if the inner c++ code woudl throw an exception lua 52 woudl usualyl just catch it
20161010 17:01:53< gfgtdf> tad_: this is not what we want speciall for thing liek quite_game_exception or network error or bad_alloc
20161010 17:02:20< tad_> I can see that. But how is that different from Lua, itself, throwing?
20161010 17:02:43< gfgtdf> how do you mean ?
20161010 17:03:14< gfgtdf> tad_: to fix this, we catch the exception at the 'lua code -> c++ code ' thne store (copy it) and then rethor it at the 'c++ code -> lua code '
20161010 17:04:00< gfgtdf> tad_: c++ doesnt have soemthign leika 'victual copy ctor' so we need a common base class for decalre a virtual 'clone' method and derive our exception from that
20161010 17:04:04< tad_> Orginally, Lua just did a catch(...) and passed it to the panic function. Was there a problem with supplying a panic function and doing the work there instead of hacking the source?
20161010 17:04:07< gfgtdf> tad_: that is lua_jailbreak_exception
20161010 17:05:20< gfgtdf> tad_: does te panic function have access to the exception object?
20161010 17:05:37 * tad_ is checking ... 1 sec
20161010 17:07:16< JyrkiVesterinen> 20161010 01:21:14< tad_> gui/widgets/listbox.cpp:334:73: warning: ‘*((void*)& horizontal_scrollbar_position +4)’ may be used uninitialized in this function
20161010 17:07:30< gfgtdf> JyrkiVesterinen: soudns familliar to me
20161010 17:07:38< gfgtdf> JyrkiVesterinen: i think this was discused yesterday
20161010 17:07:52< JyrkiVesterinen> I can't reproduce the above error. I just made an optimized GCC build with CMake and GCC 6.2.0.
20161010 17:07:54< gfgtdf> JyrkiVesterinen: ah right this is a quote
20161010 17:08:05< tad_> I was doing a first-ever SCons build (make all)
20161010 17:08:33< tad_> And it was defaults, so +O2 
20161010 17:08:38< JyrkiVesterinen> SCons-vs-CMake shouldn't matter here as long as the build has optimizations enabled.
20161010 17:09:10< JyrkiVesterinen> SCons makes optimized builds by default, CMake doesn't. Thus, I explicitly enabled optimizations when trying to get the warning myself.
20161010 17:09:11< tad_> Well, its jsut a warning so if it was missed during a build you'd never see it again until the file changed.
20161010 17:09:32< JyrkiVesterinen> I turned on strict compilation for that. Even one warning would have failed the build.
20161010 17:10:18< gfgtdf> does anyone know a good plugin for a windows textedior thatn can fold/unfold tags in wml documents? It's important that it parses the general wml syntax and not just some specific tags, (this means, it should support user defined wml tags.)
20161010 17:10:48< JyrkiVesterinen> I guess the -Wmaybe-unititialized problem is really fragile and depends on compilation order of the files.
20161010 17:10:55< JyrkiVesterinen> I plan to simply suppress it.
20161010 17:11:33< irker031> wesnoth: Gregory A Lundberg wesnoth:master 1528a8bb5a35 / src/ (9 files in 3 dirs): Refactor luaL_typerror https://github.com/wesnoth/wesnoth/commit/1528a8bb5a35c805fa9508a6f1690648f6f4a097
20161010 17:11:35< irker031> wesnoth: Gregory A Lundberg wesnoth:master 196a044dd6cf / src/lua/lgc.cpp: Revert spurious change to whitespace https://github.com/wesnoth/wesnoth/commit/196a044dd6cf0b7c56448eaeae0491cfb6a3493e
20161010 17:11:37< irker031> wesnoth: Gregory A Lundberg wesnoth:master be43e6c73091 / src/lua/llex.cpp: Fix bug: Index out of bounds https://github.com/wesnoth/wesnoth/commit/be43e6c730916e79e6861531ad38f61ef55ad59f
20161010 17:11:39< irker031> wesnoth: Gregory A Lundberg wesnoth:master 299b0cc55c70 / src/lua/lmathlib.cpp: No need to redefine PI https://github.com/wesnoth/wesnoth/commit/299b0cc55c70a8e32f3a7f0ef16d4323f063d7a4
20161010 17:11:41< irker031> wesnoth: gfgtdf wesnoth:master 426d57b4b74f / src/ (12 files in 3 dirs): Merge pull request #815 from GregoryLundberg/GL_Lua https://github.com/wesnoth/wesnoth/commit/426d57b4b74f909895ff2ce316e9bc2be9f6ebf5
20161010 17:12:09-!- Greg-Bog_ [~greg_bogg@173.240.241.83] has quit [Remote host closed the connection]
20161010 17:13:22< tad_> JyrkiVesterinen, I can try to figure out how to reproduce the warning. Might be a while, the wife is asking from help with arithmetic for a sewing project which always takes a while.
20161010 17:14:05< JyrkiVesterinen> Well, I think suppressing is the way to go anyway.
20161010 17:14:47< JyrkiVesterinen> As shadowm said, the code is correct. The warning is a false positive. GCC can't prove to itself that the value will never be accessed if it's uninitialized.
20161010 17:15:15< tad_> Suppresion generally means everywhere, so I'm resistent to that.
20161010 17:15:38< JyrkiVesterinen> Localized suppressing is perfectly possible.
20161010 17:15:39< JyrkiVesterinen> http://www.dbp-consulting.com/tutorials/SuppressingGCCWarnings.html
20161010 17:16:31< tad_> I've not looked at the code or thought about the problem. I just note and move on because I'm focused elsewhere.
20161010 17:17:44< JyrkiVesterinen> I'm the author of that code, so fixing the problems it causes is my responsibility by default.
20161010 17:20:38< tad_> Well, do what you think best. My approach is usually to honor any message from any compiler and only suppress when it can be shown to be false (or the correction is REAL gross). I consider any local suppression to be a minor code smell.
20161010 17:25:04< tad_> gfgtdf, Oh, nice. Lua source recognizes the problem and lets me do it in my new wesnothluaconfig.h file in 5.2 and 5.3 so I can simply move the hack there. Cool beans.
20161010 17:27:30-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has quit [Read error: Connection reset by peer]
20161010 17:29:27< irker031> wesnoth: Jyrki Vesterinen wesnoth:master d7bb7dc56069 / src/gui/widgets/listbox.cpp: Suppress -Wmaybe-uninitialized in tlistbox::place() https://github.com/wesnoth/wesnoth/commit/d7bb7dc560692f790653c82745e4a11021c4cc94
20161010 17:29:28-!- Greg-Boggs [~greg_bogg@173.240.241.83] has joined #wesnoth-dev
20161010 17:31:03-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has joined #wesnoth-dev
20161010 17:31:38< tad_> JyrkiVesterinen, I'd have to check the logs from last night but I seem to recall a second message out of SCons. I know I triggered one and had to fix it by how I changed the Lua library .. that may have been it.
20161010 17:32:17< JyrkiVesterinen> Was this it?
20161010 17:32:18< JyrkiVesterinen> 20161010 01:48:16< tad_> font.cpp:631:85: warning: ignoring return value of ‘FriBidiLevel fribidi_log2vis(const FriBidiChar*, FriBidiStrIndex, FriBidiParType*, FriBidiChar*, FriBidiStrIndex*, FriBidiStrIndex*, FriBidiLevel*)’, declared with attribute warn_unused_result
20161010 17:32:31< tad_> That was it.
20161010 17:32:43< JyrkiVesterinen> It's not my code, so it's not my responsibility to fix. :P
20161010 17:33:00< tad_> vultraz, ^
20161010 17:33:08< JyrkiVesterinen> And I have more stuff on my plate, so I don't intend to do anything about that warning.
20161010 17:33:47< tad_> That's where I am, too.
20161010 17:34:48< tad_> Of course, I'm sorta distacted right now what with trying to explain that I can't get a 60x60 square out of a piece of fabric which is only 43 wide.
20161010 17:40:24-!- atarocch [~atarocch@93.56.160.28] has quit [Ping timeout: 260 seconds]
20161010 17:44:50-!- Greg-Boggs [~greg_bogg@173.240.241.83] has quit [Remote host closed the connection]
20161010 17:47:52-!- travis-ci [~travis-ci@ec2-54-160-196-22.compute-1.amazonaws.com] has joined #wesnoth-dev
20161010 17:47:53< travis-ci> wesnoth/wesnoth#11424 (master - 426d57b : gfgtdf): The build passed.
20161010 17:47:54< travis-ci> Build details : https://travis-ci.org/wesnoth/wesnoth/builds/166488693
20161010 17:47:54-!- travis-ci [~travis-ci@ec2-54-160-196-22.compute-1.amazonaws.com] has left #wesnoth-dev []
20161010 17:50:29-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has quit [Ping timeout: 260 seconds]
20161010 17:51:27-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has joined #wesnoth-dev
20161010 17:52:12-!- atarocch [~atarocch@93.56.160.28] has joined #wesnoth-dev
20161010 18:06:47-!- mjs-de [~mjs-de@wh.Uni-Dortmund.DE] has joined #wesnoth-dev
20161010 18:08:02-!- Greg-Boggs [~greg_bogg@173.240.241.83] has joined #wesnoth-dev
20161010 18:21:40-!- travis-ci [~travis-ci@ec2-54-167-15-199.compute-1.amazonaws.com] has joined #wesnoth-dev
20161010 18:21:41< travis-ci> wesnoth/wesnoth#11425 (master - d7bb7dc : Jyrki Vesterinen): The build passed.
20161010 18:21:42< travis-ci> Build details : https://travis-ci.org/wesnoth/wesnoth/builds/166493137
20161010 18:21:42-!- travis-ci [~travis-ci@ec2-54-167-15-199.compute-1.amazonaws.com] has left #wesnoth-dev []
20161010 18:41:42-!- Kwandulin [~Miranda@p200300760F2C716759298D27268B3352.dip0.t-ipconnect.de] has quit [Read error: Connection reset by peer]
20161010 18:44:47-!- Greg-Boggs [~greg_bogg@173.240.241.83] has quit [Remote host closed the connection]
20161010 18:46:07-!- Greg-Boggs [~greg_bogg@173.240.241.83] has joined #wesnoth-dev
20161010 18:47:33-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Quit: Konversation terminated!]
20161010 18:49:43-!- wedge010 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has joined #wesnoth-dev
20161010 18:53:36-!- wedge009 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has quit [Ping timeout: 260 seconds]
20161010 18:53:36-!- wedge010 is now known as wedge009
20161010 19:02:26< celticminstrel> Is the lua jailbreak exception still needed? Does lua still catch(...) rather than a custom struct type?
20161010 19:04:47-!- Nobun [~nobun@5.170.107.189] has joined #wesnoth-dev
20161010 19:10:36-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev
20161010 19:13:20< tad_> celticminstrel, Yes, still needed. And easy to refactor out of the Lua souce. Check my GL_Lua2 branch if you want to see how I reverted lua/ldo.c to match released source.
20161010 19:18:35< tad_> Fun: gcc does not like memory starvation (boo, hiss Chrome you pig)
20161010 19:18:40< tad_> c++: internal compiler error: Killed (program cc1plus)
20161010 19:18:40< tad_> Please submit a full bug report,
20161010 19:18:50 * tad_ chuckles.
20161010 19:19:31-!- Duthlet [~Duthlet@dslb-188-104-253-155.188.104.pools.vodafone-ip.de] has quit [Quit: leaving]
20161010 19:20:55< JyrkiVesterinen> Hah. That problem simply doesn't exist for me. Desktop PC with maxed out RAM (32GB). :P
20161010 19:21:33< DeFender1031> JyrkiVesterinen, I'm on a desktop wth 16GB and i regularly end up in swap.
20161010 19:21:50< DeFender1031> (Yes, this thing is still configured with a swap drive)
20161010 19:22:05< JyrkiVesterinen> Heavy tab user?
20161010 19:22:22< DeFender1031> JyrkiVesterinen, very. It also usually only happens when I'm also running a vm
20161010 19:22:38< DeFender1031> been thinking of upping to 32.
20161010 19:23:16< JyrkiVesterinen> I see. I keep the number of tabs low, so I never go anywhere close of exhausting my memory.
20161010 19:23:42< DeFender1031> JyrkiVesterinen, you guys are actually part of the problem. I consistently have several dozen wesnoth reference tabs open :P
20161010 19:23:48< JyrkiVesterinen> (I decided to max out because I calculated that the price of DDR3 will probably go up at some point when manufacturing capacity is scaled down.)
20161010 19:23:57-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has quit [Ping timeout: 240 seconds]
20161010 19:24:08-!- vultraz [~chatzilla@124.109.10.167] has joined #wesnoth-dev
20161010 19:24:08-!- vultraz [~chatzilla@124.109.10.167] has quit [Changing host]
20161010 19:24:08-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has joined #wesnoth-dev
20161010 19:27:37< tad_> Well, if I can ever get the wife to stop purchasing sewing machines, sergers, and embroidery machines, maybe I can save up the cash to purchase another high-end desktop. Until the, she's appropriated my machine for her 3D online game and Facebook while I'm relegated to her old laptop.  **sighs**
20161010 19:28:51< tad_> I used to steal cycles of the big machine running my VMs over there but she insisted it slowed her game down so she booted me out.
20161010 19:33:15< tad_> My newsfeed today says Google is working to reduce the memory footpring Chrome needs. Expected for v 55, due out in a couple months unless you're running cannary
20161010 19:36:11< JyrkiVesterinen> I saw that article as well and read the actual blog post from V8 (Chrome's JavaScript engine) developers.
20161010 19:36:13< JyrkiVesterinen> http://v8project.blogspot.fi/2016/10/fall-cleaning-optimizing-v8-memory.html
20161010 19:36:31< tad_> 40% reduction sounds good
20161010 19:38:10< JyrkiVesterinen> The biggest improvement (50% lower heap size) is exclusive to the memory reduction mode... which is only used on low-memory mobile devices (~512MB).
20161010 19:38:27< JyrkiVesterinen> Improvements are much lower on PCs.
20161010 19:39:22< JyrkiVesterinen> Tech sites like to spin up the news too much...
20161010 19:39:33< celticminstrel> tad_: What does -include do?
20161010 19:39:35-!- ancestral [~ancestral@63.236.20.2] has joined #wesnoth-dev
20161010 19:39:38< celticminstrel> Is that related to PCH?
20161010 19:40:15< celticminstrel> wesnothluaconfig.h is also missing the copyright block at the top of every file.
20161010 19:40:19< tad_> celticminstrel, -include injects the file into the source as if inserting #include "<file>" at line 0.
20161010 19:40:44< tad_> celticminstrel, I'll add it.
20161010 19:40:47< celticminstrel> Has that flag been around awhile?
20161010 19:41:07< tad_> #define awhile "Since the earth cooled."
20161010 19:41:19 * celticminstrel also notes that "cannot" is one word.
20161010 19:41:29< celticminstrel> Okay, then I guess clang ~3.2 would support it.
20161010 19:41:49< celticminstrel> I know how to add flags to specific files in XCode, so I can do that part once it's merged (or send you a patchfile if you want).
20161010 19:42:14< tad_> celticminstrel, I checked. It appears it does. I should side load clang with gcc and check, I s'pose.
20161010 19:42:29-!- ancestral [~ancestral@63.236.20.2] has quit [Client Quit]
20161010 19:42:42< tad_> celticminstrel, XCode should have it to. If you can make the change and send me the patch I'll add it to the changeset.
20161010 19:43:02< celticminstrel> If I recall correctly, Lua defines a standard way to include an extra file; is there any particular reason why you didn't use a -D flag to leverage that, instead?
20161010 19:43:19< celticminstrel> I can think of one good reason to do it...
20161010 19:43:22< celticminstrel> Possibly.
20161010 19:43:57< celticminstrel> Why is there an empty wesnothluaconfig.md?
20161010 19:44:01< tad_> celticminstrel, I tried all sorts of things and this was the one which worked, did not use undocumented features on Windows.
20161010 19:44:11< tad_> celticminstrel, Because I have not written it yet.
20161010 19:44:43< celticminstrel> I think the -D flag would mean that the header is also included in anything that includes Lua headers, whereas this way means only the Lua source files themselves include it.
20161010 19:44:52< celticminstrel> Not quite sure though.
20161010 19:45:01< celticminstrel> It's been awhile since I looked at this.
20161010 19:46:07< tad_> I can -D for only the lua kit, and will if I need it.
20161010 19:46:48< tad_> The issue I had was getting my #define in before Lua looked, but at the right place to make Visual C++ and gcc both happy.
20161010 19:48:07< tad_> I spent a day working every way I could think of to avoid having to change the project files. I'm open to a better way if you can find one.
20161010 19:48:12< celticminstrel> Blah, I can't fetch from your repo because of lack of disk space. >_>
20161010 19:48:46< celticminstrel> I guess I need to clean up some stuff on my home partition.
20161010 19:49:23< tad_> Sounds like you need one of these 3T or 4T NAS boxes I have ... about USD110 at my local WallyWorld.
20161010 19:49:59< celticminstrel> I think I have like 3T total disk space (maybe a bit less) spread across three drives and about five or six partitions.
20161010 19:50:19< celticminstrel> So one way to solve this might be just moving stuff around.
20161010 19:50:34< celticminstrel> There's probably some stuff I can clear out of Downloads though, as well.
20161010 19:51:42< tad_> I remember when a huge mainframe had 64K (real 'woven' core and hand-built wire-wrapped circuits) and there wasn't 1T in the entire world. So I sorta freak when I think my wife needs that 4T just for her pictures and PDFs.
20161010 19:53:16< tad_> Back then, a 'portable' computer was one you could fit in two semi trailers. :P
20161010 19:53:17< Aginor_> tad_: that wasn't a cray was it?
20161010 19:53:51< tad_> IBM 360 Model 30 .. we needed compat mode becuase the accounting system was still written in Autocoder.
20161010 19:54:36< tad_> Cray didn't exist, yet.  But at another place I did use a CDC Cyber 6600 .. 60 bit words and it was SCREAMING fast (for the day).
20161010 19:59:15-!- stikonas_ [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev
20161010 20:00:20-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Ping timeout: 250 seconds]
20161010 20:04:04< vultraz> gfgtdf: did yous ee my question from earlier
20161010 20:05:57-!- Aginor_ is now known as Aginor
20161010 20:06:05-!- Aginor [~andreas@apollo.alternating.net] has quit [Changing host]
20161010 20:06:05-!- Aginor [~andreas@unaffiliated/aginor] has joined #wesnoth-dev
20161010 20:06:28-!- knotwork_ [~markm@99.192.88.35] has joined #wesnoth-dev
20161010 20:06:28-!- knotwork_ [~markm@99.192.88.35] has quit [Changing host]
20161010 20:06:28-!- knotwork_ [~markm@unaffiliated/knotwork] has joined #wesnoth-dev
20161010 20:07:10-!- JyrkiVesterinen [~jyrki@78-27-105-35.bb.dnainternet.fi] has quit [Quit: Going to bed]
20161010 20:07:29-!- knotwork [~markm@unaffiliated/knotwork] has quit [Ping timeout: 258 seconds]
20161010 20:07:50< tad_> celticminstrel, If Lua has a way to include a user .h file I don't see it. All I see is luaconf.h .. the only place we're supposed to change if we can't -D to select the options we need.
20161010 20:11:04< celticminstrel> I think it was in 5.3 that I remember seeing it, so perhaps it doesn't exist yet.
20161010 20:12:44< tad_> celticminstrel, I was reading the 5.3.3 source kit when I wrote that.  -include does the job, though, and seems very widely implemented. I didn't research how far back various compilers went with it .. 'far' seemed the answer, though.
20161010 20:13:17< gfgtdf> vultraz: no i didn't
20161010 20:13:41< vultraz> [00:39:02]	vultraz	gfgtdf: so, if you come back, I was trying to add entry point selection back to the mp create game flow. i thought a simple item selector dialog would work, so I copied the fetch code from the configure_engine ctor and put it in enter_configure_mode, and it works, the scenario list comes up, but if you select anything other than the first scenario, the game crashes when...
20161010 20:13:43< vultraz> ...launching the MP...
20161010 20:13:44< vultraz> [00:39:02]	vultraz	...Staging dialog. obv i need to update some setting to sync the gamestate but im not sure what
20161010 20:13:46< vultraz> [00:39:07]	vultraz	gfgtdf: commit https://github.com/Vultraz/wesnoth/commit/a5e0688525233f989576a30634024d4b7bb0ad96
20161010 20:13:57-!- knotwork__ [~markm@99.192.76.212] has joined #wesnoth-dev
20161010 20:14:20-!- knotwork_ [~markm@unaffiliated/knotwork] has quit [Ping timeout: 260 seconds]
20161010 20:26:57-!- knotwork__ [~markm@99.192.76.212] has quit [Ping timeout: 240 seconds]
20161010 20:30:35-!- irker031 [~irker@uruz.ai0867.net] has quit [Quit: transmission timeout]
20161010 20:33:59< gfgtdf> vultraz:  don't know wha it crashes but wroitign 'state.set_scenario(scenario);' there looks liek it woudl overwrite all changes previously done to the scenario like expereince_modifer stc.
20161010 20:34:37-!- Nobun [~nobun@5.170.107.189] has quit [Ping timeout: 240 seconds]
20161010 20:35:11< gfgtdf> why*
20161010 20:35:27< gfgtdf> vultraz: you have a stacktrace of the crash ?
20161010 20:36:28< shadowm> zookeeper: <shadowm> I might or might not add the ability to add custom bookmarks.
20161010 20:37:02< zookeeper> okay
20161010 20:37:36< shadowm> Drives will stay there, it's not like most people have more than a handful of them available at a time.
20161010 20:38:13-!- ChipmunkV [~vova@static-89-94-113-91.axione.abo.bbox.fr] has quit [Quit: ChipmunkV]
20161010 20:38:58< vultraz> gfgtdf: yeah the crash is in tmp_staging::update_leader_display - essentially, the first time one try to access one of the connect_engine side_engine ptrs.
20161010 20:39:18< vultraz> gfgtdf: and that doesn't happen if you just click through the dialog without changing selection
20161010 20:39:35< vultraz> gfgtdf: so what i think is that the scenario gets changed but some other setting isn't updated
20161010 20:40:04< vultraz> you're probably right about overriding settings, though, which isn't good :/
20161010 20:40:48< vultraz> optimally this would be a dropdown in Create but as you know we cannot get campaign scenario lists without reloading the config first
20161010 20:40:55< vultraz> so I
20161010 20:41:01< vultraz> 'm wondering how best to do this
20161010 20:44:38< vultraz> gfgtdf: do you know what function actually reloads the config, etc, after the dialog?
20161010 20:44:47< vultraz> prepare_for_new_level maybe?
20161010 20:45:06< vultraz> maybe i could put this in tmp_create::post_show...
20161010 20:48:24-!- mjs-de [~mjs-de@wh.Uni-Dortmund.DE] has quit [Ping timeout: 260 seconds]
20161010 20:49:43< gfgtdf> vultraz: in the forum there was some request for a campaignpart tag , if we implement this we coudl mabye use it in both sp and mp to replace startig points.
20161010 20:50:30< vultraz> (stacktrace is http://pastebin.com/K4CPS4v6 but it doesn't really say anything, essentially just confirms the crash is when the game accesses one of the side_engine ptrs)
20161010 20:51:28< vultraz> gfgtdf: well, entry points could potentially be useful to start the campaign at any point. most campaigns that would make use of the Chapters feature already feature multiple campaign entries.
20161010 20:52:18< vultraz> idk, i didn't think there was any use for them then i realized there might be some.
20161010 20:53:38-!- Nobun [~nobun@5.170.106.95] has joined #wesnoth-dev
20161010 21:07:56-!- boucman [~rosen@wesnoth/developer/boucman] has joined #wesnoth-dev
20161010 21:07:56-!- boucman [~rosen@wesnoth/developer/boucman] has quit [Read error: Connection reset by peer]
20161010 21:11:47-!- stikonas_ [~gentoo@wesnoth/translator/stikonas] has quit [Quit: Konversation terminated!]
20161010 21:12:15-!- stikonas_ [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev
20161010 21:12:45< matthiaskrgr> is minimap rendering broken?
20161010 21:14:57-!- Nobun [~nobun@5.170.106.95] has quit [Ping timeout: 240 seconds]
20161010 21:17:44< shadowm> celticminstrel: Do you know if getpwuid() is implemented on OS X and if the results are meaningful?
20161010 21:19:50 * celticminstrel runs "man getpwuid".
20161010 21:20:07< tad_> shadowm, I looked and I see comments about it having issues on Lion after a fork()
20161010 21:20:23< celticminstrel> "struct passwd* getpwuid(uid_t uid)"
20161010 21:20:36< shadowm> That answers the first part.
20161010 21:20:51< celticminstrel> "obtain information from DirectoryService(8), including records in /etc/passwd which is described in passwd(5).  "
20161010 21:21:14< shadowm> But, for example, uname is almost always implemented and the results are nigh-meaningless on Linux.
20161010 21:21:36< celticminstrel> I seem to recall using this before to get the user's login name.
20161010 21:21:40< celticminstrel> On Mac.
20161010 21:22:06< celticminstrel> Not quite sure whether it was specifically getpwuid or just one of the other functions documented on the same page.
20161010 21:26:37< vultraz> matthiaskrgr: what?
20161010 21:26:38-!- Greg-Boggs [~greg_bogg@173.240.241.83] has quit [Remote host closed the connection]
20161010 21:27:20< matthiaskrgr> minimap is empty
20161010 21:27:26< matthiaskrgr> no map previews in multiplayer screen
20161010 21:28:09-!- Greg-Boggs [~greg_bogg@173.240.241.83] has joined #wesnoth-dev
20161010 21:28:38< vultraz> cannot reproduce
20161010 21:29:04< matthiaskrgr> :o 
20161010 21:29:57< matthiaskrgr> hm
20161010 21:30:11< gfgtdf> matthiaskrgr: maybe you turned minimap off?
20161010 21:31:04< matthiaskrgr> http://imgur.com/a/tz9Xy  
20161010 21:31:28< matthiaskrgr> ok, let me look in the settings
20161010 21:31:48-!- wedge009 [~Thunderbi@60-241-236-92.static.tpgi.com.au] has quit [Ping timeout: 260 seconds]
20161010 21:32:03< gfgtdf> matthiaskrgr: there is afaik no settings in the preference for that, you cna only change this using the button below the minimap ingame
20161010 21:32:10< matthiaskrgr> oh yes :| 
20161010 21:32:36< matthiaskrgr> toggle minimap terrain draging in the editor also disables it in normal gameplay and in the multiplayer foo
20161010 21:32:39< matthiaskrgr> how weird
20161010 21:32:45< matthiaskrgr> s/draging/drawing
20161010 21:34:04< gfgtdf> vultraz: maybe w shodul ignore the 'toggle terran drawing' settign the mp create screen ?
20161010 21:34:16< matthiaskrgr> ok, sorry for false alarm then :) 
20161010 21:35:17-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has quit [Ping timeout: 240 seconds]
20161010 21:41:34-!- irker137 [~irker@uruz.ai0867.net] has joined #wesnoth-dev
20161010 21:41:34< irker137> wesnoth: Celtic Minstrel wesnoth:lua-helper-safe-metatables 79647ff9d9d9 / data/lua/helper.lua: helper.lua: set __metatable field to guard against external access to metatables https://github.com/wesnoth/wesnoth/commit/79647ff9d9d9afa6be62fc042b4d04c8c5ec5331
20161010 21:41:40-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has quit [Ping timeout: 260 seconds]
20161010 21:43:12< celticminstrel> ^ That can be merged whenever, provided that it passes and there are no syntax errors.
20161010 21:43:35< celticminstrel> It probably would fail at the WML test stage if there were syntax errors, actually.
20161010 21:45:07< celticminstrel> So it can be merged as soon as it passes, I guess.
20161010 22:20:32-!- travis-ci [~travis-ci@ec2-54-198-94-217.compute-1.amazonaws.com] has joined #wesnoth-dev
20161010 22:20:33< travis-ci> wesnoth/wesnoth#11426 (lua-helper-safe-metatables - 79647ff : Celtic Minstrel): The build passed.
20161010 22:20:33< travis-ci> Build details : https://travis-ci.org/wesnoth/wesnoth/builds/166561717
20161010 22:20:33-!- travis-ci [~travis-ci@ec2-54-198-94-217.compute-1.amazonaws.com] has left #wesnoth-dev []
20161010 22:35:12-!- horrowind [~Icedove@2a02:810a:8380:10a8:21b:fcff:fee3:c3ff] has quit [Quit: horrowind]
20161010 22:41:02-!- stikonas_ is now known as stikonas
20161010 22:45:02< tad_> Sheesh. Back in 2012 someone changed (int)luaL_checkint() to static_cast<int>(luaL_checkint()) because they didn't RTFM and it was easier to hand edit the source file instead of changing the (incorrectly writte) call site. And, now, in 2016, Lua has 'correct' the error by forcing you to do it. And all this could have been avoided by R'ing TFM.  **eyeroll**
20161010 22:51:59< shadowm> tad_: Could you elaborate?
20161010 22:54:31< tad_> luaL_checkint() is a #define alias of luaL_checkinteger() [now removed from Lua 5.3] it is simply a C (int) cast and checkinteger is already implictly convertible to int and have been since 2012.
20161010 22:55:40< tad_> I'm backing out the hand edit and finding all the places which will need fixing when I upgrade to 5.3, anyway. The Lua manual sorta says this. And I'm sure the reason the #define's are gone is exactly this mistake.
20161010 22:55:46< shadowm> The person who did this used a flag to warn on all C casts with either cppcheck or gcc itself, I don't remember. Whether the cast is redundant was probably not checked due to being beyond the scope of the check.
20161010 22:56:59< tad_> All the commit says is 'fix compiler errors'
20161010 22:57:26< shadowm> No, it says "Fix compiler warnings.", if it's the commit I'm looking at.
20161010 22:57:33< tad_> Anyway .. it will need to be done. So I'm doing it a week or two earlier.
20161010 22:57:46< tad_> shadowm, That's the one.
20161010 22:57:57< shadowm> I'm just saying this kind of thing doesn't merit this much attention.
20161010 23:00:05< tad_> It goes back to my comment about suppressing messages, I guess. Maybe back then it was not worth the attention. But the API change was probably because someone complained about having to hand-change the cast to a static_cast when compiling under C++ and the fix was to remove the unneeded cast #define
20161010 23:02:20< shadowm> The #define was there.
20161010 23:03:39< tad_> Well, I s'pose the messages gcc makes don't point out the true source of the error.
20161010 23:03:46< shadowm> AFAICT it was also used in Lua land.
20161010 23:04:02< shadowm> Lua land was compiled as C++ code.
20161010 23:04:06< shadowm> Ergo, warning.
20161010 23:04:32< shadowm> Still doesn't merit this much attention and I'm just fact-checking things.
20161010 23:04:38< tad_> I'll see. Still slogging through the Wesnoth usages.
20161010 23:05:16< tad_> So the choice for 5.3 is add the #define back in or correct the call sites. I'm correcting the call sites.
20161010 23:05:40< shadowm> The call sites are all in Lua land I believe.
20161010 23:06:13-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has joined #wesnoth-dev
20161010 23:06:17< shadowm> Okay, scratch that, they are in Wesnoth too.
20161010 23:07:49< shadowm> Anyway, the point is that if you have -Wbe-obnoxious-about-c-casts-in-cxx-code on and you have `int foo(); int bar = (int)foo();`, I suspect the compiler is still supposed to yield a warning.
20161010 23:08:48< shadowm> Since it's really a style thing to begin with and at the time the goal was to enforce certain coding conventions across the entire codebase.
20161010 23:09:52< tad_> shadowm, No the call sites are in src/scripting .. so far.
20161010 23:09:58< shadowm> Given this, accusing people of not performing an action on an F thing seems a bit excessive.
20161010 23:10:17< shadowm> src/scripting/ is Wesnoth. Lua land is src/lua/.
20161010 23:10:33< tad_> The call sites with the errors are src/scripting
20161010 23:10:43< shadowm> <shadowm> Okay, scratch that, they are in Wesnoth too.
20161010 23:12:59< tad_> It's going to have to be changed for Lua 5.3, anyway. It could have been fixed back in 2012. But, I'll admit, changing 4 lines in one place is faster than changing a few different source files.
20161010 23:13:51< shadowm> 20120205 00:06:45< vultraz> compiling on windows looks.....
20161010 23:13:51< shadowm> 20120205 00:06:48< vultraz> H A R D
20161010 23:14:10< vultraz> oh god what
20161010 23:14:10< shadowm> vultraz: I'm so proud of you. :p
20161010 23:14:24< tad_> Hmm? It's push a button and wait. And wait and wait (for me).
20161010 23:14:56< tad_> Oh, a realyl old comment.
20161010 23:17:05< vultraz> tad_: yeah, that was me four years ago :|
20161010 23:17:54< shadowm> I had a dream in which I implemented an #ifdevver WML preprocessor directive.
20161010 23:18:58< shadowm> It quickly went all wrong because right before pushing I started panicking about deciding on the exact syntax. #ifdev vs. #ifdevver vs. #if devver vs. #ifver dev and so on.
20161010 23:19:29< shadowm> I think that's the only dream I've ever had that had actual meaningful text in it.
20161010 23:19:42< tad_> Actually not a bad idea. #ifver for version number and #ifdevver for +dev on end
20161010 23:19:58< shadowm> #ifver already exists thanks to past me.
20161010 23:20:26< shadowm> Normal players and content creators have a different notion of "development" version than you appear to have.
20161010 23:21:34< shadowm> But yeah, the point is that I'm lousy at names and APIs.
20161010 23:21:50 * tad_ shrugs.
20161010 23:21:52< shadowm> I guess my subconscious is more aware of this fact than I realized.
20161010 23:22:35< vultraz> reminder than we need a #deprecated preprocessor directive
20161010 23:23:08< shadowm> Oh that's probably a task even you could accomplish unassisted.
20161010 23:23:19< tad_> So it has a different boilerplate part than #error?
20161010 23:23:38< tad_> Or so it can be suppressed with a switch?
20161010 23:23:39< shadowm> It sounds just like a glorified #warning that's able to write into the lg::wml_error or whatever stream.
20161010 23:23:49< shadowm> The one that's normally used by [deprecated_message] or what's its name.
20161010 23:24:18< vultraz> I'm busy bashing my head overt the game initialization code. 
20161010 23:24:21< shadowm> Come to think of it, maybe #warning should write into it too. :\
20161010 23:24:22< vultraz> over*
20161010 23:25:09< vultraz> none of it makes any sense
20161010 23:25:27< tad_> Let's talk about Windows and IEEE754. In Lua 5.2 we're forcing it to be 'fake' but there is no option at all like that for 5.3 .. should I expect errors?
20161010 23:25:30< shadowm> I remember messing with that like it was yesterday.
20161010 23:25:48< shadowm> I had to edit a million places and missed NULL pointer checks in most of them.
20161010 23:26:01< shadowm> And then I became a dev.
20161010 23:26:04< vultraz> "that"?
20161010 23:26:12< shadowm> MP game init code.
20161010 23:26:29< tad_> No, let's talk about rror: ‘FriBidiLevel fribidi_log2vis(const FriBidiChar*, FriBidiStrIndex, FriBidiParType*, FriBidiChar*, FriBidiStrIndex*, FriBidiStrIndex*, FriBidiLevel*)’ is deprecated [-Werror=deprecated-declarations]
20161010 23:26:41< shadowm> I assume you're talking about the code that deals with processing all the MP side setup results and setting up the initial gamestate.
20161010 23:27:17< tad_> src/font.cpp:631:85: error: ignoring return value of ‘FriBidiLevel fribidi_log2vis(const FriBidiChar*, FriBidiStrIndex, FriBidiParType*, FriBidiChar*, FriBidiStrIndex*, FriBidiStrIndex*, FriBidiLevel*)’, declared with attribute warn_unused_result [-Werror=unused-result]
20161010 23:27:25< shadowm> I know, it has changed a lot since then, but still.
20161010 23:27:31< vultraz> no, I'm dealing with the code before that
20161010 23:27:52< vultraz> the code that begins initializing a level so it can be configured for the initial gamestate :|
20161010 23:28:05< vultraz> since I want to add entry points back
20161010 23:30:45< tad_> "This function is obsolete because it only handles one-line paragraphs. Please consider using other functions instead." And no hint which.
20161010 23:38:31< vultraz> god DAMMIT.
20161010 23:39:09-!- Greg-Boggs [~greg_bogg@173.240.241.83] has quit [Remote host closed the connection]
20161010 23:39:48< vultraz> i even move this to post_show in mp create and it still chrashes
20161010 23:39:50< vultraz> *how*
20161010 23:39:53< vultraz> *HOW HOW HOW*
20161010 23:45:00< vultraz> the config is correct
20161010 23:45:08< vultraz> ats e1s3, 6 sides
20161010 23:45:19< vultraz> the connect engine has 6 side engines
20161010 23:45:22< vultraz> so what's the problem!
20161010 23:51:51< tad_> shadowm, Do all platforms have fribidi? Even Windows? Or is this error I'm getting because my system does and I should for SCons to not use it?
20161010 23:52:06< tad_> s/for SCons/force SCons/
--- Log closed Tue Oct 11 00:00:21 2016