--- Log opened Mon May 18 00:00:05 2009
--- Day changed Mon May 18 2009
20090518 00:00:05< Jetrel> add a "would be" between "space if".
20090518 00:00:07< Ivanovic> Polarina: what is said is that the overhead we currently have would be reduced *a lot* when all the images we have per unit were in a single file
20090518 00:00:09< Polarina> Jetrel: PNG files are already compressed.
20090518 00:00:18< Polarina> Ivanovic: Ok. :)
20090518 00:00:26< Ivanovic> yes, files are compressed, but all have a header
20090518 00:00:36 * Polarina would suggest MNG.
20090518 00:00:41< Blueblaze> Jetrel: check msn
20090518 00:00:42< Ivanovic> and compression of *similar stuff* is a lot better when more stuff is similar
20090518 00:00:55< alink> well, i still think that unit's sprites disk space is not important in wesnoth (terrains, portrait, mp3 use much more) memory's use otoh is important
20090518 00:00:57< Ivanovic> (and animations are very similar in each step)
20090518 00:01:07< Polarina> Ivanovic: What about using MNGs?
20090518 00:01:19 * Ivanovic does not know MNGs
20090518 00:01:24< Jetrel> Polarina:  you officially fail pragmatism forever.  We're not using a file format our artists can't even prepare images in.
20090518 00:01:35< alink> but indeed maybe there is some cpu power and IO wasted by reading all these PNG headers
20090518 00:01:36< Polarina> Ivanovic: It is a derivation of PNG, but for animated images.
20090518 00:02:16< Ivanovic> alink: i don't know how spreadsheets of images would affect ram usage
20090518 00:02:24< loonycyborg> Polarina: I once proposed using MNG too :P
20090518 00:02:28-!- grzywacz [n=grzywacz@wesnoth/developer/grzywacz] has quit [Remote closed the connection]
20090518 00:02:35< Ivanovic> that is yeah, it *can* increase usage, but it does not have to
20090518 00:02:36< Polarina> loonycyborg: And?
20090518 00:02:54< alink> Ivanovic: a simple reason is that it forces to load all frames instead of just the one we want
20090518 00:03:22< Polarina> loonycyborg: What was the response?
20090518 00:03:36< loonycyborg> Polarina: Jetrel said that it's bad idea since Photoshop doesn't support it.
20090518 00:03:36< Ivanovic> alink: implementation dependend
20090518 00:03:38< Ivanovic> ;)
20090518 00:03:47< Polarina> loonycyborg: Who the hell uses Photoshop?
20090518 00:03:53< Ivanovic> (that is: we don't have to keep everything in the image cache)
20090518 00:03:55< Jetrel> I DO.
20090518 00:04:02< Polarina> Jetrel: Use GIMP.
20090518 00:04:06< Ivanovic> Polarina: our artists
20090518 00:04:11< Ivanovic> Polarina: this is what matters
20090518 00:04:20< Jetrel> As does kitty, as does lordbob.
20090518 00:04:22< Polarina> These artists need to learn how to use real tools.
20090518 00:04:26< Jetrel> Gimp sucks.
20090518 00:04:34< Blueblaze> Jetrel: don't be mean!
20090518 00:04:34< Jetrel> I wish it didn't, I REALLY do.
20090518 00:04:35< Blueblaze> :(
20090518 00:04:37< alink> Ivanovic: So for example, users don't using idle animations, will always load them (because they are embedded in the speadsheet file)
20090518 00:04:40< Smar> gimp isn’t good at _image creation_
20090518 00:04:49< Smar> it is ok in _image manipulation_
20090518 00:05:07< Jetrel> but it's just got a really bad interface.
20090518 00:05:07< Ivanovic> alink: like i said, it depends on the implementation
20090518 00:05:17< Smar> Jetrel: no, not really
20090518 00:05:18< Ivanovic> alink: at "load time" the image could be cut up for the cache
20090518 00:05:27-!- ilor [n=user@wesnoth/developer/ilor] has quit [Read error: 60 (Operation timed out)]
20090518 00:05:28< Jetrel> Don't get me wrong - photoshop has a lot of design issues too, but gimp is even worse.
20090518 00:05:30-!- loonycyborg [n=sergey@wesnoth/developer/loonycyborg] has quit ["KVIrc 3.4.2 Shiny http://www.kvirc.net/"]
20090518 00:05:36< Ivanovic> alink: so yeah, when the image is initially loaded a little more is used, but only temporary
20090518 00:05:41 * Polarina uses Inkscape.
20090518 00:05:52-!- loonycyborg [n=sergey@79.139.246.175] has joined #wesnoth-dev
20090518 00:06:12< Ivanovic> Polarina: no way to use inkscape for pure pixel graphics
20090518 00:06:16< Ivanovic> at least efficiently
20090518 00:06:28 * loonycyborg isn't sure whether GIMP supports MNG either.
20090518 00:06:50< Jetrel> MNG is kind of a dead format - the one that's gonna succeed is APNG, anyways, because firefox supports it.
20090518 00:06:55< alink> Ivanovic: still need to read the whole file, but how do we know what images remove? for example do i remove leading anim, and when needed reload the whole spreadsheets ?
20090518 00:07:29< Ivanovic> alink: i think it might be possible to implement image spreadsheets in a way that memory usage would not be increased, i have not said that it would be trival or anything
20090518 00:07:58< Ivanovic> and yeah, spreadsheets could also only be for "animation type"
20090518 00:08:12< Ivanovic> so a unit has several spreadsheets, one for the leading anim, one for...
20090518 00:08:13< Polarina> Jetrel: I didn't know about APNG.
20090518 00:08:16-!- thespaceinvader_ [n=chatzill@91.108.140.5] has joined #wesnoth-dev
20090518 00:08:33< Ivanovic> (since only complete animations are left out normally and not partitial anims)
20090518 00:08:33< alink> Ivanovic: I agree you can make it without increasing memory, just don't cache it, and use more I/O ;-)
20090518 00:08:36-!- thespaceinvader [n=chatzill@wesnoth/artist/thespaceinvader] has quit [Nick collision from services.]
20090518 00:08:44< Jetrel> Polarina:  basically identical design intent to MNG.
20090518 00:08:54-!- thespaceinvader_ is now known as thespaceinvader
20090518 00:09:08< Ivanovic> anyway, time to head off to bed, n8
20090518 00:09:10< Jetrel> Ivanovic:  so we might think about doing that down the road.  It'd really simplify things in a  lot of areas, especially if you follow my 'holistic' suggestions for it on transparency and such.
20090518 00:09:13-!- wesbot changed the topic of #wesnoth-dev to: released 1.7.0, announcing "soon" | 60 bugs, 241 feature requests, 10 patches | logs: http://irclogs.wesnoth.org/ | Don't paste on IRC! Use a pastebin: http://wesnoth.pastebin.com | http://imagebin.org
20090518 00:09:34< alink> gn Ivanovic
20090518 00:09:38< Ivanovic> Jetrel: to be honest: i prefer to leave the *main* decision with this to the artists
20090518 00:09:38< Jetrel> probably want to wait quite a while before we do though.
20090518 00:09:46-!- crimson_penguin [n=ben@wesnoth/developer/crimsonpenguin] has quit []
20090518 00:09:52< Ivanovic> since artists have to work with the stuff and be compftable with it
20090518 00:10:12< Jetrel> Ivanovic:  perhaps in several months.
20090518 00:10:22< Jetrel> Ivanovic:  thanks.  :)
20090518 00:10:27< Polarina> Jetrel: Wikipedia says MNG is more powerful than APNG.
20090518 00:10:51< Ivanovic> "wikipedia says that pidgeons live in oceans"
20090518 00:10:53< Ivanovic> ;)
20090518 00:11:02< alink> Jetrel: if it for artists, maybe a script to separate and join these images when needed? 
20090518 00:11:14-!- stikonas [n=and@wesnoth/translator/stikonas] has quit [Read error: 110 (Connection timed out)]
20090518 00:11:19< alink> s/it/it is
20090518 00:11:52< alink> ah no, windows artists too :-/
20090518 00:12:25< Jetrel> alink:  I won't ever need a script to separate them.   I'm now used to editing them in sheets.
20090518 00:13:09< alink> Jetrel: how do you encode what frame is where? in a separate cfg file? is it handy too ?
20090518 00:13:41< Jetrel> alink:  basically, yes - each frame in a config file has it's rectangle specified.
20090518 00:13:48 * alink will go check frogatto code :)
20090518 00:14:16< Jetrel> alink:  actually frogatto eliminated the need to specify individual frames for most animations.  It assumed that we actually don't need to edit the individual frame timings.
20090518 00:14:45< Jetrel> alink:  I didn't think that would be a good assumption, but frogatto demonstrates strongly that it's actually just fine.
20090518 00:15:18< Jetrel> In the rare case you actually need to have separate timings, you just string multiple animations together.
20090518 00:16:38< alink> there is certainly good points to do it like that (after all a lot of sprite games do like this), but some details still annoys me a little. I will check frogatto code and rethink about this 
20090518 00:16:39< Jetrel> alink:  an entire animation can be specified in some 10 lines of FML;  it's ...  really an improvement.
20090518 00:17:23< loonycyborg> Jetrel: Is Photoshop ported to Mac or you use some form of virtualization to run it?
20090518 00:17:27< Jetrel> alink:  we probably -don't- want to overhaul wesnoth's image system like this - we probably want to keep our current WML and just have rectangle specifiers in the image-path.
20090518 00:18:00< Jetrel> loonycyborg:  actually, photoshop is ported to windows.  It was originally mac-only.
20090518 00:18:29< loonycyborg> Fascinating :P
20090518 00:18:41< Jetrel> loonycyborg:  and there are quite a few mac-only programs around these days, several of which are in my toolchain, such as GraphicConverter.
20090518 00:19:09< Jetrel> loonycyborg:  what's even more fascinating is a growing mac-only OSS community. :)
20090518 00:20:28< Jetrel> It doesn't have to be mac-only; the core APIs were an open standard in OPENStep, but it hasn't really caught on.
20090518 00:22:02 * loonycyborg is very unlikely to become a member of mac-only OSS community, since even if he somehow gets a Mac he'll install Gentoo on it :)
20090518 00:23:01< loonycyborg> AFAIK there are non-mac implementations of OPENStep.
20090518 00:23:44< Jetrel> loonycyborg:  yeah - GNUStep is one, as is a windows-based work-in-progress called Cocoatron.
20090518 00:25:11< Jetrel> Eventually, as macs get more and more market share, this'll probably provide a much nicer cross-platform migration path than the various attempts to clone the win32 api.
20090518 00:25:27-!- Espreon [n=espreon@wesnoth/developer/espreon] has quit ["Tengo que ir... Yeahzorz..."]
20090518 00:26:14< Polarina> Wine!
20090518 00:26:42< Jetrel> Both macs and linux are on the rise at the expense of windows; right now the former is at ~10%, and the latter at 1%.
20090518 00:28:05< Jetrel> Anyways.
20090518 00:38:24< CIA-30> alink * r35721 /trunk/src/display.cpp: Small simplification/optimization of propagate_invalidation()
20090518 00:40:39< CIA-30> alink * r35722 /trunk/src/unit_animation.cpp: use STL function when possible
20090518 00:47:15< loonycyborg> I've found a bug. If in test scenario you summon a troll, move him twice and then dismiss him, then you'll get an assert.
20090518 00:47:44< loonycyborg> wesnoth: src/actions.cpp:2579: void apply_shroud_changes(undo_list&, game_display*, const gamemap&, unit_map&, std::vector<team, std::allocator<team> >&, int): Assertion `unit_itor != units.end()' failed.
20090518 00:49:02< alink> loonycyborg: indeed, I investigate
20090518 00:49:57< alink> mmmh the WML in the 'dismiss' part is less my area, but maybe not related to that
20090518 00:53:20-!- ilor_ [n=user@wesnoth/developer/ilor] has quit [Remote closed the connection]
20090518 00:54:21-!- ilor [n=user@wesnoth/developer/ilor] has joined #wesnoth-dev
20090518 00:59:02< CIA-30> alink * r35723 /trunk/src/ (display.cpp game_display.cpp unit_frame.cpp unit_frame.hpp): Use reference for image::locator (not so small structure)
20090518 01:08:50< alink> loonycyborg: ok got it, easy to do a quick fix, a bit more work to fix the global same flaw also present in others place
20090518 01:09:10< alink> * other places
20090518 01:10:36< alink> note that you can already avoid the crash by clearing your undo stack before dismissing, for example by doing another move clearing shroud
20090518 01:19:58< CIA-30> alink * r35724 /trunk/src/actions.cpp: 
20090518 01:19:59< CIA-30> Fix assert-crash spotted by loonycyborg in test-scenario
20090518 01:19:59< CIA-30> (dismissing troll after a move leaving the undo-stack not empty)
20090518 01:19:59< CIA-30> Current fix has the small flaw that a unit WML-killed by an event will not
20090518 01:19:59< CIA-30> clear the shroud explored by its previous move (if delay shroud is on)
20090518 01:20:19< alink> but maybe can be done by wml
20090518 01:21:53-!- ilor_ [n=user@wesnoth/developer/ilor] has joined #wesnoth-dev
20090518 01:21:59< alink> the perfect fix will need code to be able to know *before* firing the event if it will need to clear the undo. Because when do it after the event, some info (a killed unit) are missing to update the delayed shroud (which uses undo)
20090518 01:25:15-!- Fab1 [n=Fab@ANantes-552-1-58-200.w92-139.abo.wanadoo.fr] has quit [Read error: 104 (Connection reset by peer)]
20090518 01:26:34-!- BenUrban_ [n=benurban@unaffiliated/benurban] has joined #wesnoth-dev
20090518 01:26:40-!- noy [n=Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20090518 01:28:14< loonycyborg> Actually, I just wanted to check that lua works in 1.7.0 release. But having done that I've decided to explore the test scenario further while I'm at it :P
20090518 01:29:40< alink> strange that i didn't noticed it before, I summon trolls there everyday :-p
20090518 01:30:11-!- thespaceinvader [n=chatzill@wesnoth/artist/thespaceinvader] has quit ["ChatZilla 0.9.84 [Firefox 3.0.10/2009042316]"]
20090518 01:30:57-!- ABCD [n=ABCD@wikipedia/ABCD] has joined #wesnoth-dev
20090518 01:33:00< alink> there is maybe another possible fix. I just found that a copy of the unit was stored in the undo stack, but need unit_map voodoo to temporary replace this ghost on the map
20090518 01:34:38-!- ilor [n=user@wesnoth/developer/ilor] has quit [Read error: 110 (Connection timed out)]
20090518 01:38:28-!- noy [n=Noy@wesnoth/developer/noy] has quit []
20090518 01:40:08< alink> good, that worked. But need some polishing and details checking. I will do that and commit later.
20090518 01:43:44-!- Chusslove [n=caslav@brsg-d9beffb7.pool.mediaWays.net] has quit [Read error: 110 (Connection timed out)]
20090518 01:44:55-!- BenUrban [n=benurban@unaffiliated/benurban] has quit [Read error: 110 (Connection timed out)]
20090518 01:45:21-!- BenUrban_ is now known as BenUrban
20090518 01:45:41< Polarina> Checking for C header file sys/sendfile.h... yes
20090518 01:45:41< Polarina> Checking for C function sendfile()... no
20090518 01:45:41< Polarina> Why?
20090518 01:46:06-!- Blueblaze [n=nick@c-98-199-143-139.hsd1.tx.comcast.net] has quit [Remote closed the connection]
20090518 01:47:49< alink> grrr stupid gna page caching, sorry ilor_ for the bug update spam
20090518 01:50:14< Polarina> Anyone?
20090518 01:51:45< loonycyborg> Polarina: Perhaps you look at build/config.log?
20090518 01:53:43< CIA-30> jetryl * r35725 /trunk/data/core/images/units/orcs/ (16 files): Cleanup on grunt and warrior images.
20090518 01:54:17< Polarina> loonycyborg: I forgot to install boost.
20090518 01:56:12< loonycyborg> I do know why sendfile() check fails if iostreams check fails.
20090518 01:57:01< Polarina> Compile warning with g++-4.4 on branches/1.6: http://rafb.net/p/gqvVcK23.html
20090518 01:57:54< Soliton> upgrade to 4.5.
20090518 01:58:00< Polarina> http://rafb.net/p/kx7Bet49.html another.
20090518 01:58:56< CIA-30> jetryl * r35726 /trunk/data/core/images/units/orcs/ (7 files): Visual overhaul for the orc leader, and added a new leadership anim macro.
20090518 01:59:23< Polarina> Breaking strict-aliasing rules is bad for the optimizer.
20090518 01:59:37< CIA-30> jetryl * r35727 /trunk/data/core/ (15 files in 3 dirs): Visual overhaul for the orc leader, and added a new leadership anim macro.
20090518 02:00:51< Polarina> I am off to bed now.
20090518 02:00:56-!- Polarina [n=polarina@wesnoth/translator/Polarina] has quit ["Leaving."]
20090518 02:03:39< CIA-30> jetryl * r35728 /trunk/data/core/ (22 files in 2 dirs): Visual overhaul for the orc ruler.
20090518 02:03:40-!- loonycyborg [n=sergey@wesnoth/developer/loonycyborg] has quit ["Zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz"]
20090518 02:04:15-!- loonybot [n=loonybot@wesnoth/bot/loonybot] has quit [Remote closed the connection]
20090518 02:06:11< CIA-30> jetryl * r35729 /trunk/data/core/ (22 files in 2 dirs): Visual overhaul for the orc sovereign.
20090518 02:08:40-!- chains [n=Rylar@adsl-69-209-64-30.dsl.chcgil.sbcglobal.net] has joined #wesnoth-dev
20090518 02:36:31-!- Elvish_Pillage2 [n=eli@68-116-175-24.dhcp.oxfr.ma.charter.com] has quit ["Hi! I'm a quit message virus vaccine. If you see a quit message virus, don't replace your quit message with it!"]
20090518 02:55:22-!- [Relic] [n=[Relic]@adsl-76-230-212-46.dsl.milwwi.sbcglobal.net] has joined #wesnoth-dev
20090518 02:55:28-!- alink [n=alink@wesnoth/developer/alink] has quit [Remote closed the connection]
20090518 02:56:24< [Relic]> Hello :)
20090518 02:59:56-!- Jetrel [n=Jetrel@wesnoth/artist/jetrel] has left #wesnoth-dev []
20090518 03:26:52-!- ilor [n=user@wesnoth/developer/ilor] has joined #wesnoth-dev
20090518 03:31:01-!- ilor_ [n=user@wesnoth/developer/ilor] has quit [Read error: 60 (Operation timed out)]
20090518 03:39:39-!- ancestral [n=ancestra@97-116-120-23.mpls.qwest.net] has joined #wesnoth-dev
20090518 04:05:00-!- Ivanovic_ [n=ivanovic@dtmd-4db2bad8.pool.einsundeins.de] has joined #wesnoth-dev
20090518 04:06:28-!- Blueblaze [n=nick@c-98-199-143-139.hsd1.tx.comcast.net] has joined #wesnoth-dev
20090518 04:20:27-!- Ivanovic [n=ivanovic@wesnoth/developer/ivanovic] has quit [Read error: 113 (No route to host)]
20090518 04:20:59-!- Ivanovic_ is now known as Ivanovic
20090518 05:03:03-!- ancestral [n=ancestra@97-116-120-23.mpls.qwest.net] has quit []
20090518 05:10:21-!- crimson_penguin [n=ben@64.201.60.216] has joined #wesnoth-dev
20090518 05:30:38-!- Appleman1234 [n=Appleman@131.181.47.17] has joined #wesnoth-dev
20090518 05:43:36-!- chains [n=Rylar@adsl-69-209-64-30.dsl.chcgil.sbcglobal.net] has quit [Remote closed the connection]
20090518 05:58:48-!- ancestral [n=ancestra@97-116-120-23.mpls.qwest.net] has joined #wesnoth-dev
20090518 05:58:50-!- ancestral [n=ancestra@97-116-120-23.mpls.qwest.net] has quit [Remote closed the connection]
20090518 06:00:00-!- crimson_penguin [n=ben@wesnoth/developer/crimsonpenguin] has quit []
20090518 07:00:36-!- silene [n=plouf@ASte-Genev-Bois-152-1-32-34.w83-114.abo.wanadoo.fr] has joined #wesnoth-dev
20090518 07:10:22-!- Appleman1234 [n=Appleman@131.181.47.17] has quit [Read error: 110 (Connection timed out)]
20090518 07:12:07-!- Sirp [n=me@wesnoth/developer/dave] has quit ["leaving"]
20090518 07:42:24-!- maxy [n=maxy@80-219-0-199.dclient.hispeed.ch] has joined #wesnoth-dev
20090518 08:01:40-!- [Relic] [n=[Relic]@adsl-76-230-212-46.dsl.milwwi.sbcglobal.net] has quit ["Leaving"]
20090518 08:10:29-!- YogiHH [i=d4ca9d15@wesnoth/developer/yogihh] has joined #wesnoth-dev
20090518 08:37:43-!- Appleman1234 [n=Appleman@131.181.47.17] has joined #wesnoth-dev
20090518 08:49:38-!- stikonas [n=and@ctv-213-164-125-83.vinita.lt] has joined #wesnoth-dev
20090518 09:18:31-!- Blueblaze [n=nick@c-98-199-143-139.hsd1.tx.comcast.net] has quit [Remote closed the connection]
20090518 09:24:28-!- euschn [n=chatzill@wesnoth/developer/euschn] has joined #wesnoth-dev
20090518 09:24:33< euschn> hi
20090518 09:25:02< YogiHH> hi euschn
20090518 09:25:45-!- cjhopman [n=chris@wesnoth/developer/cjhopman] has joined #wesnoth-dev
20090518 09:30:06-!- stikonas [n=and@wesnoth/translator/stikonas] has quit [Read error: 104 (Connection reset by peer)]
20090518 09:31:21-!- stikonas [n=and@ctv-213-164-125-83.vinita.lt] has joined #wesnoth-dev
20090518 09:31:42-!- silene [n=plouf@wesnoth/developer/silene] has quit [Read error: 60 (Operation timed out)]
20090518 09:33:55-!- stikonas [n=and@wesnoth/translator/stikonas] has quit [Remote closed the connection]
20090518 09:39:03-!- cjhopman [n=chris@wesnoth/developer/cjhopman] has quit [Remote closed the connection]
20090518 09:59:53-!- busfahrer [n=busfahre@unixboard/user/busfahrer] has joined #wesnoth-dev
20090518 10:08:01-!- loonybot [n=loonybot@79.139.246.175] has joined #wesnoth-dev
20090518 10:18:54-!- zookeeper [n=l@wesnoth/developer/zookeeper] has joined #wesnoth-dev
20090518 10:43:30< Soliton> looks like we have to make campaignd ping stale connections some time. atm the 1.4 and 1.6 campaignd have more established connections each than the 1.6 wesnothd. :-P
20090518 10:45:02< CIA-30> ilor * r35730 /trunk/src/server/ (game.cpp game.hpp player_network.cpp player_network.hpp): extract group data sending functions from wesnothd's game class to free functions in a separete header/cpp pair
20090518 10:45:19< CIA-30> ilor * r35731 /trunk/ (4 files in 2 dirs): add new file to build systems
20090518 10:45:38< CIA-30> ilor * r35732 /trunk/src/server/ (game.cpp player_network.cpp player_network.hpp): fix and use the predicate version of send_to_many
20090518 10:45:57< CIA-30> ilor * r35733 /trunk/src/server/server.cpp: fix unused variable warning on win32
20090518 11:02:44-!- Elvish_Pillager [n=eli@68-116-175-24.dhcp.oxfr.ma.charter.com] has joined #wesnoth-dev
20090518 11:06:06-!- EdB [n=edb@136.117.88-79.rev.gaoland.net] has joined #wesnoth-dev
20090518 11:22:00-!- loonycyborg [n=sergey@79.139.246.175] has joined #wesnoth-dev
20090518 11:28:39-!- busfahrer [n=busfahre@unixboard/user/busfahrer] has quit ["leaving"]
20090518 11:36:33< Rhonda> Hmm, http://bugs.debian.org/529170 makes me remember some discussion from in here from shortly after 1.6 release?
20090518 11:36:55< Rhonda> I think people said that there is no proof, but if different people report it I think there might be something to it ...
20090518 11:43:25< ilor> I think there was an issue with animations being internally processed during ai planning, but that was fixed
20090518 11:45:25< Rhonda> In what version?
20090518 11:46:51< ilor> not sure, I think 1.6
20090518 11:46:56< ilor> fixed in 1.6.1
20090518 11:49:05< Rhonda> Well, the bugreport mentions 1.6.2 as still affected by quite some ai-time related cpu usage
20090518 12:00:34-!- Appleman1234 [n=Appleman@131.181.47.17] has quit [Read error: 110 (Connection timed out)]
20090518 12:05:21-!- Crab_ [n=Crab_@wesnoth/developer/crab] has joined #wesnoth-dev
20090518 12:05:27< Crab_> hi
20090518 12:07:55< Crab_> Rhonda: some issues with animations were fixed only a few days ago, and some - only in trunk, not in branches/1.6
20090518 12:09:13-!- wesbot changed the topic of #wesnoth-dev to: released 1.7.0, announcing "soon" | 60 bugs, 242 feature requests, 10 patches | logs: http://irclogs.wesnoth.org/ | Don't paste on IRC! Use a pastebin: http://wesnoth.pastebin.com | http://imagebin.org
20090518 12:17:57-!- EdB [n=edb@136.117.88-79.rev.gaoland.net] has quit [Remote closed the connection]
20090518 12:18:29< Rhonda> Crab_: I'm not totally convinced that this really should be related. animations also happen during player turns.
20090518 12:19:35< Crab_> Rhonda: yes. but, the underlying problem is that human *thinks* in parallel with animations, but the ai *thinks* only between animations.
20090518 12:19:59< Crab_> so, especially on slow computers, most of the ai turn time is spent on animating stuff.
20090518 12:21:24< Crab_> especially on maps with lots of units.
20090518 12:33:19< Rhonda> Crab_: Not sure wether "Kernel: Linux 2.6.26-1-686 (SMP w/2 CPU cores)" could be considered a slow computer.
20090518 12:34:21< Crab_> 'SMP w/2 CPU cores' is irrevelant.  and the kernel version says little about the computer :)
20090518 12:35:45< Crab_> for example, on my computer (2GHz Pentium M), on my test case, we were having 33 minutes per ai sides turn, and then boucman & silene & alink & me decreased that to 9.5 minutes using various optimizations.
20090518 12:38:17< Rhonda> ouch, 33 minutes?!?
20090518 12:38:33< Rhonda> Was it that bad with 1.4? Is that really only fixed in trunk?
20090518 12:38:39< Crab_> Rhonda: yes. 7 minutes of actual turn and 26 minutes of animations.
20090518 12:38:47< Crab_> Rhonda: trunk and 1.7.0
20090518 12:38:51< Crab_> Rhonda: some when into 1.6.2
20090518 12:39:26< Rhonda> I'll ask for the savegame and relevant side-informations about what scenario/map this happens with, then.
20090518 12:39:27< Crab_> and my test case is pathological (big map, lots of animated units). although it is from mainline campaign, Northern Rebirth.
20090518 12:40:26< Crab_> and note that, after the improvements, we had "~10 minutes on 2000 MHz, ~60 minutes on 600 MHz". So, there's non-liner slowdown on slower computers.
20090518 12:40:35< Crab_> s/non-liner/non-linear
20090518 12:40:58-!- Polarina [n=polarina@wesnoth/translator/Polarina] has joined #wesnoth-dev
20090518 12:41:42< Crab_> afaik, in 1.4 it was faster (the animation code was less powerful)
20090518 12:41:42< wesbot> Crab_: Sometimes we are fast
20090518 12:41:57 * Rhonda bites wesbot
20090518 12:42:30< Crab_> is "Sometimes we are fast" copyrighted by wesbot :) ?
20090518 12:43:38< ilor> anyone familiar with network_worker.cpp around?
20090518 12:44:05< Crab_> Rhonda: a quick way to check if GUI is the cause, is to run the ai turn minimized
20090518 12:44:50< Crab_> or 'with all animations and sounds disabled'
20090518 12:45:50< Crab_> for example, for one poster on the forum, 'disabling all animations and sounds' decreased the time of ai turn (5 ai sides) on that big map from 12+ hours to 2-3 hours.
20090518 12:47:41< Rhonda> I think I should give NR a turn, which scenario is that? number, not name. :)
20090518 12:48:00< Crab_> 13a_Showdown.cfg
20090518 12:49:06< Crab_> You need to have a good number of units (~100 units in sum for all human players, and let the orcs ai recruit for a few turns, too)
20090518 12:53:00< Rhonda> And propably most of those 100 units to level 2 or 3?
20090518 12:53:14 * Polarina thinks wesnoth is loading the PNG animation files over and over each step.
20090518 12:53:21< Rhonda> Is that some sort of epic scenario?  *gulp*
20090518 13:03:52-!- ilor_ [n=user@wesnoth/developer/ilor] has joined #wesnoth-dev
20090518 13:07:38-!- ilor [n=user@wesnoth/developer/ilor] has quit [Read error: 60 (Operation timed out)]
20090518 13:17:01< Crab_> Rhonda: yes, that's the 'epic battle' scenario :)
20090518 13:21:43-!- jmunro[a] is now known as jmunro
20090518 13:39:23-!- stikonas [n=and@ctv-213-164-125-83.vinita.lt] has joined #wesnoth-dev
20090518 14:42:14-!- ardesh [n=ardesh@port-92-206-6-74.dynamic.qsc.de] has joined #wesnoth-dev
20090518 14:52:45-!- alink [n=alink@wesnoth/developer/alink] has joined #wesnoth-dev
20090518 14:54:03-!- ilor [n=user@wesnoth/developer/ilor] has joined #wesnoth-dev
20090518 14:57:13-!- ilor_ [n=user@wesnoth/developer/ilor] has quit [Read error: 60 (Operation timed out)]
20090518 14:57:20< Crab_> Rhonda: and, regarding AI speed in general, you can see http://wesnoth.terraninfo.net/?dim1=svn_release for stats 'ai-vs-ai without gui on a small map', and note how ai became faster in recent svn (mainly, thanks to pathfinding improvements)
20090518 15:18:04-!- BenUrban [n=benurban@unaffiliated/benurban] has quit ["Power failu"]
20090518 15:21:03-!- mjs-de [n=mjs-de@vpw.wh.Uni-Dortmund.DE] has joined #wesnoth-dev
20090518 15:28:55-!- crimson_penguin [n=ben@64.201.60.216] has joined #wesnoth-dev
20090518 15:47:36-!- Elvish_Pillage2 [n=eli@68-116-175-24.dhcp.oxfr.ma.charter.com] has joined #wesnoth-dev
20090518 15:49:10-!- Elvish_Pillager [n=eli@68-116-175-24.dhcp.oxfr.ma.charter.com] has quit [Read error: 110 (Connection timed out)]
20090518 15:51:01-!- Polarina [n=polarina@wesnoth/translator/Polarina] has quit [Read error: 104 (Connection reset by peer)]
20090518 15:54:34-!- busfahrer [n=busfahre@unixboard/user/busfahrer] has joined #wesnoth-dev
20090518 15:59:04-!- Turuk [n=Turuk@wesnoth/forumsith/turuk] has joined #wesnoth-dev
20090518 16:01:08-!- Polarina [n=polarina@wesnoth/translator/Polarina] has joined #wesnoth-dev
20090518 16:09:14-!- ilor_ [n=user@wesnoth/developer/ilor] has joined #wesnoth-dev
20090518 16:20:32-!- fendrin [n=fabi@wesnoth/developer/fendrin] has joined #wesnoth-dev
20090518 16:27:40-!- ilor [n=user@wesnoth/developer/ilor] has quit [Read error: 110 (Connection timed out)]
20090518 16:28:23< crimson_penguin> I figured out what the difference between opening Wesnoth from XCode vs. the Finder was - I had the Finder set to open the PPC version (meaning it had to do emulation)
20090518 16:28:41< crimson_penguin> It's interesting though, that with PPC there's all these graphics problems
20090518 16:29:17-!- Polarina [n=polarina@wesnoth/translator/Polarina] has quit [Read error: 110 (Connection timed out)]
20090518 16:37:54-!- Polarina [n=polarina@wesnoth/translator/Polarina] has joined #wesnoth-dev
20090518 16:49:44-!- Turuk [n=Turuk@wesnoth/forumsith/turuk] has quit ["The man who smiles when things go wrong has thought of someone to blame it on."]
20090518 16:57:48-!- euschn [n=chatzill@wesnoth/developer/euschn] has quit ["ChatZilla 0.9.84 [Iceweasel 3.0.6/2009032803]"]
20090518 16:58:47-!- Chusslove [n=caslav@brsg-d9beec0c.pool.mediaWays.net] has joined #wesnoth-dev
20090518 16:59:29< Chusslove> What is the meaning of @-sign followed by speaker's ID in some WML strings? E.g.
20090518 16:59:32< Chusslove> story= _ "@Kalenz
20090518 16:59:44< Chusslove> and then empty line followed by character's lines.
20090518 17:03:40-!- Netsplit verne.freenode.net <-> irc.freenode.net quits: Alesis-Novik, Soliton
20090518 17:04:23-!- Netsplit over, joins: Soliton, Alesis-Novik
20090518 17:05:06-!- Polarina [n=polarina@wesnoth/translator/Polarina] has quit [Read error: 110 (Connection timed out)]
20090518 17:09:17< Chusslove> Nah, scratch my last question.
20090518 17:09:42-!- Polarina [n=polarina@wesnoth/translator/Polarina] has joined #wesnoth-dev
20090518 17:13:44-!- stikonas [n=and@wesnoth/translator/stikonas] has quit [Read error: 60 (Operation timed out)]
20090518 17:28:41< fendrin> Chusslove: I tried to bring wml to colour the speakers name red with the @.
20090518 17:29:21< Chusslove> fendrin: Thanks, seen it after remembering that there is actually the code I could look into :)
20090518 17:29:28< Soliton> not a very good attempt since that's green. :-P
20090518 17:31:30-!- Blueblaze [n=nick@c-98-199-143-139.hsd1.tx.comcast.net] has joined #wesnoth-dev
20090518 17:33:48-!- stikonas [n=and@wesnoth/translator/stikonas] has joined #wesnoth-dev
20090518 17:36:01-!- busfahrer [n=busfahre@unixboard/user/busfahrer] has quit ["leaving"]
20090518 17:38:36< fendrin> Soliton, Chusslove: It didn't work. I will try to do it with the new highlighting syntay that was introduced lately pangosomething
20090518 17:39:47< CIA-30> ilor * r35734 /trunk/src/network_worker.cpp: 
20090518 17:39:47< CIA-30> Fix a networking problem when trying to send data to a disconnected socket
20090518 17:39:47< CIA-30> According to SDL docs, SDLNet_TCP_Send is blocking and if the returned
20090518 17:39:47< CIA-30> nubmer of bytes sent is less than expected, it's an error. Previous code
20090518 17:39:47< CIA-30> could trigger an infinite loop with the function returning zero and our
20090518 17:39:47< CIA-30> loop trying to send again and again.
20090518 17:40:06< CIA-30> ilor * r35735 /trunk/projectfiles/VC9/wesnoth.sln: VC9 wesnothd solution settongs update
20090518 17:48:45-!- ShikadiLord [n=ignacio@wesnoth/developer/shadowmaster] has joined #wesnoth-dev
20090518 17:49:45-!- ShikadiLord [n=ignacio@wesnoth/developer/shadowmaster] has quit [Nick collision from services.]
20090518 17:49:47< ilor_> how do I backport a commit in git?
20090518 17:51:00< loonycyborg> ilor_: git cherry-pick
20090518 17:55:26< CIA-30> ilor * r35736 /branches/1.6/src/network_worker.cpp: 
20090518 17:55:26< CIA-30> Fix a networking problem when trying to send data to a disconnected socket
20090518 17:55:26< CIA-30> According to SDL docs, SDLNet_TCP_Send is blocking and if the returned
20090518 17:55:26< CIA-30> nubmer of bytes sent is less than expected, it's an error. Previous code
20090518 17:55:26< CIA-30> could trigger an infinite loop with the function returning zero and our
20090518 17:55:27< CIA-30> loop trying to send again and again.
20090518 17:56:45< ilor_> I've to admit, git is really nice
20090518 17:58:42< Ivanovic> hi
20090518 17:58:53< fendrin> hi Ivanovic
20090518 17:59:35< Ivanovic> fendrin: hmm, there was something i wanted from you some days ago...
20090518 17:59:52< Ivanovic> ah, right, probably this report: https://gna.org/bugs/index.php?13511
20090518 17:59:54< fendrin> Ivanovic: maybe a fix for 1.7 low chief must die?
20090518 18:00:10< fendrin> Ivanovic: juat spotted it.
20090518 18:00:15< fendrin> s/juat/just
20090518 18:01:31< Ivanovic> yes, this should be fixed in 1.7
20090518 18:01:40< Ivanovic> (in current trunk that is, will be in 1.7.1)
20090518 18:01:58< fendrin> Ivanovic: That behaviour is new. The ai must have changed. In 1.6 the wolves don't activate every unit on the screen. Maybe dragonking modified the formula or the formula backend was changed.
20090518 18:02:28< Ivanovic> i don't know, i just know that esrs wife encountered this problem
20090518 18:06:57< fendrin> Ivanovic: I will investigate it this weekend.
20090518 18:07:18< Ivanovic> good
20090518 18:08:39< fendrin> wesbot: seen dragonking?
20090518 18:08:39< wesbot> fendrin: The person with the nick Dragonking last spoke 7d 22h ago.  1d 15h ago was here and on the channels #wesnoth and #wesnoth-umc-dev with the message:
20090518 18:09:13-!- wesbot changed the topic of #wesnoth-dev to: released 1.7.0, announcing "soon" | 59 bugs, 242 feature requests, 10 patches | logs: http://irclogs.wesnoth.org/ | Don't paste on IRC! Use a pastebin: http://wesnoth.pastebin.com | http://imagebin.org
20090518 18:09:35< fendrin> Is there a possibility to search for all commits done by dragonking?
20090518 18:10:45< Soliton> git shortlog --author=dragonking :->
20090518 18:11:35< Crab_> fendrin: and what is wanted from these wolves ?
20090518 18:12:48-!- YogiHH [i=d4ca9d15@wesnoth/developer/yogihh] has left #wesnoth-dev []
20090518 18:12:53< fendrin> Crab_: I believe dragonking wanted the current behaviour, it looks intented..
20090518 18:13:36< fendrin> Crab_: He may have enhanced the patrol code to wake the surounding units but that leads to the need of rebalancing the scenario.
20090518 18:14:46< Crab_> well, the last time he touched the patrol formula was "Mar 18". is the current behavior that old ?
20090518 18:15:55< fendrin> Crab_: I see. Was that before the split to 1.6?
20090518 18:16:18< Crab_> yes. but you can change the wolves 'smell radius' to be smaller...
20090518 18:16:50< Crab_> it's " if( distance_between( closest_unit(ai, me).loc, me.loc ) > me.movement_left-1," line
20090518 18:17:01< fendrin> Crab_: Yeah, but if it isn't a change in the patrol formula ai code it's a bug in wesnoth's ai.
20090518 18:17:29< Crab_> possible, too. or, it may be that it WAS bug in wesnoth ai, but now it's fixed :)
20090518 18:17:37< fendrin> The formula ai is resetting the ai=guardian parameter for some reason to default ai.
20090518 18:18:16< fendrin> I wouldn't have thought that this is even possible.
20090518 18:18:58< Crab_> I'll load that scenario now and take a look...
20090518 18:19:15< Crab_> what's the difficulty level ?
20090518 18:20:40< fendrin> Crab_: Feel free to do so. But I can't stay with you. There is a software finish deadline for me this midnight. Some nasty kernel hacking for a operating systems lecture. Disabling virtual interrupts and other evil low level stuff in c without ++.
20090518 18:20:56< Crab_> ok. and say what should be the correct behavior
20090518 18:21:26< Crab_> a) "wolves should not not wake up friendly units" ?
20090518 18:21:35-!- jmunro is now known as jmunro[a]
20090518 18:21:51< fendrin> Crab_: I am not even sure. The new behaviour looks like a feature for me. It could make the scenario more interresting if balanced well. That is why I thougt someone would have coded that willingly.
20090518 18:21:52< Crab_> or b) "wolves should detect you on closer distance then their movement range-1"
20090518 18:23:23< fendrin> Crab_: First try to see if the new behaviour is a bug.
20090518 18:24:26< Soliton> pretty difficult to detect buggy behaviour if you have no idea what the intended behaviour is.
20090518 18:24:48< Crab_> ok, if one of the wolves detects you, then all orcs attack you. that seems ok for me - it's perfectly understandable (alert raised)
20090518 18:26:16< fendrin> Crab_: can't find "(alert raised)" in the code.
20090518 18:26:32< Crab_> that's not in the code, that just my thoughts
20090518 18:26:59< fendrin> Crab_: yeah, mine as well. But I don't remember me coding it.
20090518 18:27:47< Crab_> ok, the orcs are missing guardian states from the start
20090518 18:28:27< Crab_> is this a bug ?
20090518 18:28:31< ardesh> question: do you have some good links to wml coding?
20090518 18:28:40< ardesh> mostly tutorials
20090518 18:29:01< ardesh> and how to do "deeper" changes to the game itself with wml
20090518 18:29:27< Soliton> wiki.wesnoth.org/ReferenceWML
20090518 18:29:41< zookeeper> umm...i'm looking at 16_The_Chief_must_die.cfg and nowhere does it seem to tell the player about wolves being able to sniff out the player.
20090518 18:30:01< zookeeper> oh, right, the invisible tooltip.
20090518 18:30:33< zookeeper> "if close enough" is completely vague though, the player should be told the exact range
20090518 18:30:44< fendrin> zookeeper:  message= _ "This feels passing strange! Landar, remember to stay far from the wolves-they may be able to smell us."
20090518 18:30:53 * zookeeper missed that one too
20090518 18:31:32< zookeeper> anyways, AFAICT to be successful you have to know how large that range is at least approximately, yet the player isn't told that.
20090518 18:31:46< ardesh> Soliton, thy
20090518 18:32:26 * zookeeper is afk now
20090518 18:32:33< fendrin> Crab_: the guard radius is set to 3 for most wolves.
20090518 18:33:57< Crab_> fendrin: so, they should detect a player if they pass within 3 squares of player ?
20090518 18:34:21< fendrin> Crab_: Yeah, I guess that is the meaning of guard_radius.
20090518 18:35:41< fendrin> pulling enmies in that range shouldn't be that evil either. I guess that will fix the scenario for now.
20090518 18:38:44< fendrin> Crab_: Sorry, I have to leave now. Thank you :-)
20090518 18:38:51-!- fendrin [n=fabi@wesnoth/developer/fendrin] has quit [Remote closed the connection]
20090518 18:48:25-!- Noyga [n=lame-z@wesnoth/developer/noyga] has joined #wesnoth-dev
20090518 18:53:27< Crab_> fendrin: ok, I see. well, the map is definitely beatable, but to do it one the 1st try and without reloading, the player will need to 1) know the guard radius (as zookeeper has said) 
20090518 18:53:40< Crab_>  2) sit for first 7 turns doing nothing but calculating the riders' waypoints (as the player cannot set labels on the map on enemy turn, he'll need to use a pencil-and-paper approach to write their movements, and set labels later). and then, after patrol routes are known,  the player will need to calculate the 'safe path' and stick to it.
20090518 18:55:28< Crab_> the wolves are too fast - so it's very hard to 'predict' their movements, and to 'predict' safe zones (without labeling them beforehand) , so trying to beat the map rush-style will lead to many reloads.
20090518 18:56:59< Crab_> I've got a suggestion: spare the player from that hard work of labeling patrol routes (or remembering them in some other way), (at least on easy/normal difficulties) - place the labels "Wolf X was here on turn Y" when the wolves move each turn.
20090518 18:57:58< Crab_> then, any player who's patient enough will have the routes after those 7 turns (7 is longest patrol route), will still have a puzzle to solve "how to get to that chief?"
20090518 18:58:08< Crab_> and no save\reloads will be required.
20090518 18:59:25< Soliton> maybe you can make the routes fairly obvious via map design?
20090518 18:59:48< Crab_> Soliton: hard, since each wolf uses his own route, and they're chaotic enough
20090518 19:00:14< Crab_> from the story perspective, incremental 'labeling of patrol routes' this will simulate the fact that Kalenz is smart enough (and is patient enough) to remember all those wolves movements :)
20090518 19:03:26< Crab_> 12 wolves, each with ~9 moves, are too fast for 38x30 map :)
20090518 19:04:19< Crab_> and note that the wolf hasn't got to 'end a move' within 3 squares of the elf - he can detect on 'passing by' and attack instead
20090518 19:06:09< Crab_> (ok, I go now, will return later today)
20090518 19:06:20-!- Crab_ [n=Crab_@wesnoth/developer/crab] has quit ["Leaving."]
20090518 19:07:05< Soliton> placing labels seems a bit out of character to me but with dialog from Kalenz that sounds good.
20090518 19:11:05-!- boucman [n=rosen@wesnoth/developer/boucman] has joined #wesnoth-dev
20090518 19:13:09< boucman> hey all
20090518 19:13:50< boucman> wesbot: seen Crab_
20090518 19:13:50< wesbot> boucman: The person with the nick Crab_ last spoke 7m 41s ago.  7m 30s ago they left with the message: "Leaving."
20090518 19:14:00< boucman> damn
20090518 19:14:20< alink> hi boucman
20090518 19:14:53< alink> boucman: just found a need_update problem but hard to track
20090518 19:15:06< boucman> i'm studying the last diff between silen's change and mine, and at this point I think his patch is faster for Crab's special case, but might be slower otherwise
20090518 19:15:43< alink> sadly, this bug has probably annoying perf problem :(
20090518 19:16:38< alink> boucman: after some animation (at least idle and select), some standing units always report need_update =true
20090518 19:17:16< alink> and thus always invalidate their hexes
20090518 19:17:34< boucman> is it easy to reproduce ?
20090518 19:17:39< boucman> I'll check that in a moment
20090518 19:18:04< alink> yeah, but i try with test scenario to have a clear example
20090518 19:18:44< boucman> thx
20090518 19:18:48< boucman> compiling right now
20090518 19:19:22< alink> use :sunset and just select deselect units, they stay stuck in invalidation loop
20090518 19:19:35< alink> (= stay bright in sunset)
20090518 19:19:40< boucman> k, will check
20090518 19:20:03-!- Polarina1 [n=polarina@85-220-26-48.dsl.dynamic.simnet.is] has joined #wesnoth-dev
20090518 19:20:11< alink> and I noticed that r35577 used different need_update functions
20090518 19:20:43< boucman> wesbot: log r35577
20090518 19:20:44< wesbot> boucman * r35577 : various animation optimisation (cache result of overlap calculation + better initialization of image::locator)
20090518 19:20:47< wesbot> URL: http://svn.gna.org/viewcvs/wesnoth?view=rev&rev=35577
20090518 19:21:09< boucman> alink: theoretically, we use a different need_update only if not on screen
20090518 19:23:31< alink> unit_frame::need_update didn't seem to check anim time, only does_not_change, but i believe that unit::animation::need_update() is more exact and use time, so any badly updated anim will cause problem
20090518 19:24:22< boucman> I don't understand what you mean
20090518 19:24:25< alink> (maybe I am completely wrong, the object hierarchy is not fully clear in my mind)
20090518 19:24:45< boucman> its : animation->particle->frame
20090518 19:26:00< alink> boucman: ok forget about that for the moment, just check if you see the source of the bug, and i do a svn search of the commit causing it :)
20090518 19:26:14< boucman> ok, 
20090518 19:26:17-!- Polarina [n=polarina@wesnoth/translator/Polarina] has quit [Read error: 60 (Operation timed out)]
20090518 19:26:34< boucman> though, it's probably 35577
20090518 19:29:16-!- Polarina1 is now known as Polarina
20090518 19:31:09< alink> boucman: compiling r35576... in the mean time, i see that the old unit_frame::invalidate called the simple unit_frame::need_update to decide if it must invalidate things. But now we call need_update in unit_animation, and unit_animation::need_update seems more advanced
20090518 19:32:00-!- esr [n=chatzill@static-71-162-243-5.phlapa.fios.verizon.net] has quit [Remote closed the connection]
20090518 19:32:09< alink> it call animated<unit_frame>::need_update()  and i assume that it's different than unit_frame::need_update()
20090518 19:32:24< alink> animated use time stuff etc
20090518 19:34:55< boucman> unit_aimation::need_update callse the need_update for all unit_animatio::particule::need_update (which makes sense)
20090518 19:35:32-!- mordante [n=mordante@wesnoth/developer/mordante] has joined #wesnoth-dev
20090518 19:35:47< mordante> servus
20090518 19:35:59< boucman> alink: i'm a bit confused with what you said
20090518 19:36:09< boucman> there is no more unit_frame::invalidate IIRC
20090518 19:36:15< alink> boucman: yeah it make sense, just pointing that we didn't do like that before, so maybe it show only now a bug like a missing update_last_draw_time or something like that
20090518 19:36:26< alink> hi mordante
20090518 19:36:30< boucman> possible
20090518 19:36:31< ilor_> hi mordante
20090518 19:36:42< mordante> hi alink, ilor_ 
20090518 19:37:01< alink> boucman: to be clear: the current way seems good, but trigger a new bug
20090518 19:37:20< boucman> ok :)
20090518 19:37:36< Ivanovic> hi mordante
20090518 19:37:42< alink> boucman: looking what changed may help to find what trigger this new bug
20090518 19:37:49< Ivanovic> mordante: you saw the probs you created yesterday in the logs?
20090518 19:38:05< boucman> indeed
20090518 19:38:08< alink> boucman: r35576 doesn't have it, trying r35577
20090518 19:38:23< boucman> but my computer is acting weird right now, I'll look at it asap
20090518 19:38:27< ilor_> mordante: as you might have noticed, I ditched the player group class in favor of some free functions for sending data to many players. looks simpler, I'll be working on the actual "room" class today
20090518 19:38:30< mordante> hi Ivanovic no, but reading the log
20090518 19:40:13< Ivanovic> hehe, should i mark this one as "invalid" since we already have patches (xdelta)? https://gna.org/bugs/index.php?13528
20090518 19:40:16< mordante> Ivanovic, hmm I'm quite sure I reverted game.cpp before committing (seems I failed to do that properly)
20090518 19:40:51< mordante> ilor_, cool, I'll look at your patches after finishing the log
20090518 19:40:54-!- silene [n=plouf@wesnoth/developer/silene] has joined #wesnoth-dev
20090518 19:41:00< silene> hi
20090518 19:41:33< alink> hello silene
20090518 19:41:39< silene> ilor_: i'm confused by your patch to network_worker.cpp; you have made ~40 lines dead code; this can't be right
20090518 19:42:02< mordante> hi silene 
20090518 19:44:21-!- noy [n=Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20090518 19:45:15< ilor_> silene: the sdl doc is rather clear, if the send function returns less than len it's an error
20090518 19:45:32< ilor_> silene: and I had wesnothd looping endlessly there
20090518 19:45:36< silene> ilor_: yes, but you have killed all the existing error handling code
20090518 19:46:10< silene> please remove it from the file, since it can no longer be executed, you have put a "return" right before it
20090518 19:46:42< silene> i'm not saying you are wrong, but you have made a mess of the function
20090518 19:47:00< Ivanovic> loonycyborg: IIRC there was some talk about patches for the windows binary
20090518 19:47:08< alink> boucman: strange r35577 is bug-free too, i continue to search
20090518 19:47:10< Ivanovic> what was the result of the discussion?
20090518 19:47:42< boucman> r35577 was my last commit and I havn't seen anything invalidation related since...
20090518 19:48:45< ilor_> silene: the code looks like it's designed for a non-blocking send that can return after sending part of the buffer
20090518 19:48:59< mordante> Polarina, http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40156
20090518 19:49:14-!- giusef [n=giusef@unaffiliated/giusef] has joined #wesnoth-dev
20090518 19:49:24< ilor_> and yeah I got carried away and didn't notice this if/return effectively disables the code below
20090518 19:50:35< alink> wesbot: log 35659
20090518 19:50:36< wesbot> boucman * r35659 : commit all my pending perf improvements, there is too much stuff lying around I need to save all that
20090518 19:50:39< wesbot> URL: http://svn.gna.org/viewcvs/wesnoth?view=rev&rev=35659
20090518 19:51:01< Polarina> mordante: Ok.
20090518 19:51:04< boucman> alink: my bad, I thought that was commit 35577
20090518 19:51:08< alink> boucman: I am currently checking that ^
20090518 19:51:20< boucman> that one is a likely culprit
20090518 19:52:13< loonycyborg> Ivanovic: IIRC author of that idea placed his work somewhere on the wiki.
20090518 19:58:00< alink> boucman: r35659 causes it
20090518 19:58:12< boucman> not really a suprise
20090518 20:00:36< boucman> alink: you see this problem only with standing and idle anims ?
20090518 20:00:43< boucman> no other animss (that's a big clue for me)
20090518 20:00:45< ilor_> silene: I'll get to it in a sec, need to go afk for a while now
20090518 20:01:11< alink> boucman: indeed I try to check other anim too, that was just an easy test
20090518 20:03:15< boucman> ok, tell me when you know
20090518 20:04:07< alink> seems to only happen after an standing following an idle or a select
20090518 20:04:41< boucman> hmm
20090518 20:05:19< boucman> alink: in unit_animation::need_minimal_update, could you try disabling the play_offscreen_ test and tell me if it solves the problem ?
20090518 20:06:03-!- VladimirSlavik [n=chatzill@wesnoth/translator/VladimirSlavik] has joined #wesnoth-dev
20090518 20:07:13< alink> no it didn't
20090518 20:07:24< boucman> hmm
20090518 20:07:36< alink> but replacing if(need_update()) by if (false) fix it
20090518 20:07:52< boucman> where did you do that ?
20090518 20:07:55< alink> meaning that it seems caused by need_update
20090518 20:08:04< alink> in unit_animation::invalidate
20090518 20:08:09-!- grzywacz [n=grzywacz@wesnoth/developer/grzywacz] has joined #wesnoth-dev
20090518 20:08:28< boucman> k
20090518 20:08:44< boucman> so need_update seems a bit overenthousiastic
20090518 20:11:57< boucman> that doesn't make much sense to me at this point
20090518 20:12:25< alink> yes me too, that's mainly why i asked you ;-p
20090518 20:14:07< boucman> ok, another test for you :)
20090518 20:14:54-!- silene [n=plouf@wesnoth/developer/silene] has quit ["Leaving."]
20090518 20:15:11< boucman> could you replace the unit::refresh   in unit.hpp with http://wesnoth.pastebin.com/m5bafac76
20090518 20:16:06< alink> gbrmlblm touching hpp , but ok i try
20090518 20:16:25-!- mjs-de [n=mjs-de@vpw.wh.Uni-Dortmund.DE] has quit [Read error: 113 (No route to host)]
20090518 20:16:47< boucman> sorry about that...
20090518 20:16:58< boucman> you can de-inline it for testing, maybe
20090518 20:17:03< boucman> it'll do less recompile
20090518 20:17:52< alink> before that, seems that need_update=always true is caused by if(parameters_.need_update()) return true;
20090518 20:18:21< boucman> ok, cool
20090518 20:18:27-!- mjs-de [n=mjs-de@wh.uni-dortmund.de] has joined #wesnoth-dev
20090518 20:18:32< boucman> that's a huge clue, no need to do the hpp change
20090518 20:18:56< alink> ok cool too :)
20090518 20:19:18-!- noy [n=Noy@wesnoth/developer/noy] has quit []
20090518 20:20:15-!- YogiHH [n=chatzill@c223147.adsl.hansenet.de] has joined #wesnoth-dev
20090518 20:21:03-!- VladimirSlavik [n=chatzill@wesnoth/translator/VladimirSlavik] has left #wesnoth-dev []
20090518 20:21:27< boucman> alink: can you determine which part of parameters_ said it needed update ?
20090518 20:22:36< alink> mmh sorry, seems that i was too quick to accuse it
20090518 20:22:48< boucman> k
20090518 20:23:02< alink> at least two of the 3 need_update seems to always return true
20090518 20:23:07< boucman> that's a bit reassuring, I couldn't see how parameters_ could cause that
20090518 20:24:07< alink> still, parameters_.need_update() seems to return true, but another one does it too
20090518 20:24:53< boucman> alink: you sure of the unit you are looking at ?
20090518 20:25:02< boucman> sometime, it's supposed to return true :)
20090518 20:26:04< alink> btw what is this animated<unit_frame>::need_update(), a static thing ?
20090518 20:27:17< boucman> well, a particule is an animated<unit_frame> (from an inheritence point of view)
20090518 20:27:28< boucman> this call will return true if the frame has changed
20090518 20:27:47< Ivanovic> Soliton: around?
20090518 20:27:59< Ivanovic> the server currently experiences *REALLY* high mysql load
20090518 20:28:14< Ivanovic> (182% cpu load)
20090518 20:28:28< alink> ok, seems to happen for all unit: both animated<unit_frame>::need_update() and parameters_.need_update() always return true
20090518 20:28:44< boucman> that makes sense,
20090518 20:28:55< boucman> ...or not
20090518 20:29:14< alink> testing with a && to be sure
20090518 20:29:33< Soliton> Ivanovic: you can look at /var/log/mysql/mysql-slow.log for the slwo query info.
20090518 20:29:38< boucman> just add a printf testing all these at the beginning of the function, I guess
20090518 20:29:45-!- esr [n=chatzill@static-71-162-243-5.phlapa.fios.verizon.net] has joined #wesnoth-dev
20090518 20:30:17< Ivanovic> Soliton: have you already turned off the stats.wesnoth.org webpages (by deactivating the apache redirect)?
20090518 20:30:33< Soliton> Ivanovic: nope.
20090518 20:30:53< alink> replacing the body of unit_animation::particule::need_update()   by    if(parameters_.need_update() && animated<unit_frame>::need_update()) return true   triggers the bug
20090518 20:31:46< Soliton> Ivanovic: ah, crawl.yahoo.net found the stats page.
20090518 20:32:07-!- Crab_ [n=Crab_@wesnoth/developer/crab] has joined #wesnoth-dev
20090518 20:32:24< boucman> alink: what unit are you testing with ?
20090518 20:32:30< Crab_> hi boucman
20090518 20:32:38< boucman> hey Crab_
20090518 20:32:44-!- Polarina1 [n=polarina@85-220-26-48.dsl.dynamic.simnet.is] has joined #wesnoth-dev
20090518 20:33:05< alink> boucman: several of test-senario, I will test all of them. but at least peasant in NR had it too
20090518 20:33:15< boucman> k
20090518 20:33:23< Ivanovic> AUTSCH
20090518 20:33:35< boucman> I can't understand why parameters_ would return true...
20090518 20:33:46< boucman> I'd be really interested if you could find out why
20090518 20:34:31< Ivanovic> crimson_penguin: are you sure that the uploaded mac binary works?
20090518 20:34:46< Ivanovic> i fear that it might be broken since the size *decreased* compared to 1.6.2
20090518 20:35:00< Ivanovic> instead of increasing several MB
20090518 20:35:42< alink> boucman: i will try your refresh change, if idle or select(which change highlight) stay blocked at the end and no standing is started, I suppose that it could cause such behavior
20090518 20:36:20< boucman> indeed, that's one of the few areas where idle is different from other animations
20090518 20:40:06-!- Polarina [n=polarina@wesnoth/translator/Polarina] has quit [Nick collision from services.]
20090518 20:40:09-!- Polarina1 is now known as Polarina
20090518 20:40:17< alink> btw, i used select for the previous debug tests, the bug is also there for idle, but some debug tests were possibly different
20090518 20:40:23< Polarina> Please use KiB, MiB, GiB, etc. when referring to storage size of base 1024 instead of 1000. Thank you. :)
20090518 20:41:44< Ivanovic> Polarina: i use what i use, that is basically my decision...
20090518 20:41:47< crimson_penguin> Ivanovic: I'll check things
20090518 20:41:52< Ivanovic> crimson_penguin: great
20090518 20:42:45< crimson_penguin> Ivanovic: well, it is fully uploaded at least
20090518 20:43:01< crimson_penguin> I mean, I didn't check the online size, but it said it was 100% uploaded, and the binary IS smaller
20090518 20:44:19< crimson_penguin> huh... well somehow the MP server doesn't work, but it IS included, so that's not decreasing the size
20090518 20:45:24< Ivanovic> something is probably missing
20090518 20:45:27< alink> boucman: ok the refresh change fixed it,  I bet that beginning refresh by "if (state_ != STATE_STANDING) return;" prevent other state to return to standing when finished
20090518 20:45:28< Ivanovic> are all the libs included?
20090518 20:45:44< Ivanovic> any other mac users around that could help us check if the mac binary really works?
20090518 20:45:53< boucman> alink: something like that, I rushed that last change a bit, I'll review it asap
20090518 20:46:02< Ivanovic> what is really strange is that it is smaller than the windows installer
20090518 20:46:07< Ivanovic> which is unrealistic
20090518 20:46:28< Ivanovic> (simply because two binaries are shipped, ppc and x86 with all the required libs)
20090518 20:46:36< Ivanovic> crimson_penguin: are translations included?
20090518 20:46:44< Ivanovic> is the new campaign included, is the new music included, ...
20090518 20:46:46< crimson_penguin> Ivanovic: :O
20090518 20:46:52< crimson_penguin> damnit! I knew I forgot something :(
20090518 20:47:00< crimson_penguin> it's the translations
20090518 20:47:37 * crimson_penguin forgot all about them, and also forgot to read his list of things to do to package...
20090518 20:47:46< Ivanovic> argh
20090518 20:47:49< Ivanovic> not good...
20090518 20:47:51< Ivanovic> ;)
20090518 20:48:01< alink> boucman: ok i leave it to you, bbl for test if needed
20090518 20:48:02< Soliton> maybe turn the list into a script. :-)
20090518 20:48:08< boucman> k
20090518 20:48:17< boucman> could you just tell me how to reproduce ?
20090518 20:48:41< crimson_penguin> Soliton: well some of it isn't quite scriptable
20090518 20:48:50< crimson_penguin> I guess the translations should be though
20090518 20:49:11< boucman> alink: ^^^
20090518 20:49:20< alink> boucman: ":sunset" select-deselect one or few units, they stay bright (instaed of darkening like others)
20090518 20:49:29< boucman> k, testing
20090518 20:49:38< alink> disable idle to avoid the bug to randomly hit too 
20090518 20:49:44< boucman> ok
20090518 20:51:58< crimson_penguin> well this is no good to have (even in an ifdef APPLE) in data/game_config.cfg: wesnothd_name="/Applications/Battle for Wesnoth 1.5/Wesnoth MP Server.app/Contents/MacOS/Wesnoth MP Server"
20090518 20:53:00< crimson_penguin> it's not even using that though, so I'm wondering where it's getting it
20090518 20:53:40< Soliton> can you make that version independent?
20090518 20:54:21< crimson_penguin> well it's not just version dependent, it's location dependent
20090518 20:55:36< crimson_penguin> it's not looking there anyway though: sh: /Users/ben/projects/wesnoth-1.7/MacOSX/build/Release/Wesnoth.app/Contents/MacOS//wesnothd: No such file or directory
20090518 20:56:02< alink> boucman: better test (avoid sunset) change the last frame of idle to something visibly different compared to standing, observe that the anim stay stuck on that frame
20090518 20:56:29< boucman> alink: that's ok, I've reproduced, and I think I understand now, should be fixed pretty fast
20090518 20:57:08< alink> btw maybe animated<T,T_void_value>::need_update() shouldn't return true when the anim reached the end ?
20090518 20:57:29-!- Sir_Konrad [n=tjb@dpclt034064.direcpc.com] has joined #wesnoth-dev
20090518 20:57:32< alink> (assuming it's part of this bug)
20090518 20:57:58< boucman> it's not part of the bug
20090518 20:58:07< boucman> and no, we don't need update so to speek
20090518 20:58:18< boucman> i.e the image displayed has not change
20090518 20:58:41< boucman> what we (might) need is a new animation, but that's a completely different encapsulation level
20090518 20:59:11-!- stikonas [n=and@wesnoth/translator/stikonas] has quit [Read error: 54 (Connection reset by peer)]
20090518 20:59:15< alink> I thought that the ended idle/select anim continued to return need_update = true because of that
20090518 20:59:46-!- stikonas [n=and@ctv-213-164-125-83.vinita.lt] has joined #wesnoth-dev
20090518 20:59:56< boucman> i'm still investigating right now
20090518 21:00:43< alink> and the previous reset to standing when idle/select is ended prevented that "bug" to triggers before
20090518 21:01:36< alink> anyway, too hungry, really off now. bbl
20090518 21:01:45< boucman> alink: you might be right actually
20090518 21:02:43-!- silene [n=plouf@wesnoth/developer/silene] has joined #wesnoth-dev
20090518 21:03:58< crimson_penguin> I'm trying to not check for lua in configure, and I even tried --disable-game (since that's all that uses it), but to no avail
20090518 21:04:10< crimson_penguin> I just want it to make the Makefile for po/
20090518 21:04:24< boucman> Crab_: how is it going with the AI ?
20090518 21:05:13-!- Sir_Konrad [n=tjb@dpclt034064.direcpc.com] has left #wesnoth-dev []
20090518 21:05:59< silene> crimson_penguin: you have to comment configure.ac:844 if you want to disable lua checking
20090518 21:06:09< crimson_penguin> gah
20090518 21:06:44< Crab_> boucman: well, yesterday I've seen that I'll not be able to get everything ready in time for 1.7.0, and now I've got two large chunks of uncommitted code 1) ai contexts rework (will need some merging), 2) composite ai with c++ engine and c++ rca loop.
20090518 21:07:26< boucman> ok, could you possibly commit some intermediary step for me to test, or is it too unstable for that ?
20090518 21:08:00< Crab_> boucman: 1st part is fairly stable, and I'll commit it soon
20090518 21:08:23< Crab_> boucman: 2nd part *worked*, but will need some merging
20090518 21:08:30< crimson_penguin> hah! I set the environment variables, that worked
20090518 21:08:39-!- Elvish_Pillage2 [n=eli@68-116-175-24.dhcp.oxfr.ma.charter.com] has quit [Read error: 110 (Connection timed out)]
20090518 21:08:48-!- Elvish_Pillager [n=eli@68-116-175-221.dhcp.oxfr.ma.charter.com] has joined #wesnoth-dev
20090518 21:09:07< Crab_> boucman: I'm also wondering how's dragonking going  -     to create a fai engine for my composite ai, I need to change-or-wrap his candidate actions stuff
20090518 21:09:45< crimson_penguin> Ivanovic: /sw/bin/msgfmt: fr.po: headerfield `PO-Revision-Date' missing in header
20090518 21:09:57< boucman> Crab_: I don't know, I havn't talked to him for a while
20090518 21:10:04< Ivanovic> crimson_penguin: smack Noyga and silene
20090518 21:10:20< crimson_penguin> Ivanovic: then what? :P
20090518 21:10:27 * crimson_penguin smacks Noyga and silene 
20090518 21:10:33< Ivanovic> crimson_penguin: then it will magically work
20090518 21:10:33< boucman> Crab_: if the changes are not smeantic (i.e only tweaking how it's called, not the game logic) you can probably commit them
20090518 21:10:55 * crimson_penguin is pretty sure it's still broken :-/
20090518 21:12:04< Crab_> boucman: flow control will be altered, and class hierarchy will be altered, too. I'll see if I can get away with adaptors
20090518 21:13:58< Ivanovic> crimson_penguin: it is only fixed they react, too
20090518 21:14:15< silene> crimson_penguin: which file is it? (msgfmt doesn't complain here)
20090518 21:14:28-!- YogiHH [n=chatzill@c223147.adsl.hansenet.de] has left #wesnoth-dev []
20090518 21:14:41< crimson_penguin> looks like sotbe
20090518 21:14:47< Ivanovic> hmm, so what i should write about in the release announcement?
20090518 21:14:48< Crab_> silene: question about r35715: why "int" for 'side number' instead of "unsigned int" or some typedef'ed int or unsigned int ?
20090518 21:15:40< silene> Crab_: the side number needs neither the extra range nor the modulo arithmetic, so there is no need to not use the fastest integer type
20090518 21:15:59< Crab_> ok, understood.
20090518 21:20:18< crimson_penguin> ok, rather than try to change where it looks for wesnothd, I'm just gonna change the name of wesnothd (to wesnothd) so it finds it without modification
20090518 21:22:37< Ivanovic> 1.7.0 forum text: http://www.wesnoth.org/forum/viewtopic.php?f=5&t=25506
20090518 21:22:37-!- Polarina [n=polarina@wesnoth/translator/Polarina] has quit [Read error: 104 (Connection reset by peer)]
20090518 21:23:24< Ivanovic> and the frontpage is updated, too
20090518 21:23:40-!- Ivanovic changed the topic of #wesnoth-dev to: 59 bugs, 242 feature requests, 10 patches | logs: http://irclogs.wesnoth.org/ | Don't paste on IRC! Use a pastebin: http://wesnoth.pastebin.com | http://imagebin.org
20090518 21:23:41< Crab_> boucman: for example, see the difference in rca loops: http://wesnoth.pastebin.com/m9b417ea
20090518 21:24:03< Ivanovic> and yes, i list 1.7.0 on the frontpage right from the start, this is intentional
20090518 21:24:47< Ivanovic> crimson_penguin: once the mac binary is updated please ping me and i will alter the frontpage (please update the downloads page yourself)
20090518 21:25:20< crimson_penguin> Ivanovic: ok, this'll take at least 3 hours once I've got it made
20090518 21:25:31< crimson_penguin> Ivanovic: I'm not sure what to do about the po for now though
20090518 21:26:02< crimson_penguin> heh, changing the name of wesnothd made it think it needed to rebuild EVERYTHING
20090518 21:26:04< boucman> Crab_: in what way does that affect DK ?
20090518 21:26:47-!- ShikadiLord [n=ignacio@wesnoth/developer/shadowmaster] has joined #wesnoth-dev
20090518 21:26:59-!- ShikadiLord [n=ignacio@wesnoth/developer/shadowmaster] has quit [Nick collision from services.]
20090518 21:28:30< Crab_> to make his candidate actions play nice with my rca loop, I need to reorganize some things in formula ai. For example, fai candidate actions do not know fully how to execute themselves - they need some preparation before running. also, .fai needs some invalidation after every move, regardless of whether this is a fai move or not.
20090518 21:28:50< boucman> k
20090518 21:29:21< Crab_> my best bet is an adaptor (wrap his actions in my interface), but I don't know if DK has got some uncommitted changes atm which totally change fai rca
20090518 21:33:48< Crab_> boucman: also, the way .fai config is parsed, should be refactored a bit, to allow parsing of individual formula snippets
20090518 21:34:33< CIA-30> ivanovic * r35737 /trunk/po/wesnoth-sotbe/fr.po: fixed missing ?\194?\187PO-Revision-Date?\194?\171 in french translation file
20090518 21:34:37< Ivanovic> crimson_penguin: okay, fixed this in the french file
20090518 21:34:51< Ivanovic> though it should *NOT* be a fatal error, basically only some kind of warning
20090518 21:34:58< crimson_penguin> Ivanovic: I guess I can just stick this in there? "PO-Revision-Date: 2009-03-06 08:00+0200\n"
20090518 21:35:03< crimson_penguin> yeah
20090518 21:35:14-!- Polarina [n=polarina@wesnoth/translator/Polarina] has joined #wesnoth-dev
20090518 21:36:03< Ivanovic> jupp
20090518 21:36:51-!- maxy [n=maxy@80-219-0-199.dclient.hispeed.ch] has quit []
20090518 21:40:14-!- Polarina1 [n=polarina@85-220-26-48.dsl.dynamic.simnet.is] has joined #wesnoth-dev
20090518 21:40:31< crimson_penguin> Ivanovic: I'm downloading the gettext source, see if I can build/install it myself
20090518 21:42:40-!- Blueblaze [n=nick@c-98-199-143-139.hsd1.tx.comcast.net] has quit [Remote closed the connection]
20090518 21:46:21-!- boucman [n=rosen@wesnoth/developer/boucman] has quit [Read error: 104 (Connection reset by peer)]
20090518 21:46:41-!- boucman [n=rosen@wesnoth/developer/boucman] has joined #wesnoth-dev
20090518 21:49:36< ilor_> gaaah
20090518 21:50:01< ilor_> I found another infinite loop caused by ignoring the error return value from a sdl function, this time it's receive
20090518 21:50:29-!- CheeseLord [n=chatzill@78-105-212-150.zone3.bethere.co.uk] has joined #wesnoth-dev
20090518 21:50:37< ilor_> this time it explicitly states "ignore the error" at network_worker.cpp:239
20090518 21:51:05-!- CheeseLord [n=chatzill@78-105-212-150.zone3.bethere.co.uk] has left #wesnoth-dev []
20090518 21:51:10< ilor_> Soliton around? blame points to you
20090518 21:51:47< Soliton> pretty sure you can dig deeper than me.
20090518 21:52:47< ilor_> sld doc says if the return value is <0, it's an error and the socket is bad
20090518 21:52:52-!- boucman [n=rosen@wesnoth/developer/boucman] has quit ["Leaving."]
20090518 21:52:59-!- Elvish_Pillage2 [n=eli@68-116-175-221.dhcp.oxfr.ma.charter.com] has joined #wesnoth-dev
20090518 21:56:09< Soliton> ilor_: not that SDL is not necessarily called.
20090518 21:56:22< Soliton> s/not/note/
20090518 21:56:37< ilor_> ah, very true in this case
20090518 21:57:16< Soliton> infact the official server uses raw sockets there.
20090518 21:57:50< ilor_> still, recv returning <0 is an error
20090518 21:59:24< Soliton> ok, pretty sure the server also does not use that if(true) code path so do as you see fit. i don't remember what the deal was.
20090518 21:59:41< silene> ilor_: note that eagain is not an error
20090518 21:59:50-!- Polarina [n=polarina@wesnoth/translator/Polarina] has quit [Read error: 110 (Connection timed out)]
20090518 21:59:57< ilor_> I can't yet grasp what is the reason for this bit of code
20090518 22:00:22-!- boucman [n=rosen@wesnoth/developer/boucman] has joined #wesnoth-dev
20090518 22:01:52< Soliton> it waits for more data?
20090518 22:02:26< ilor_> is EWOULDBLOCK defined on linux (or on the official server)
20090518 22:03:00< silene> ilor_: no, but eagain is, and it is actually the value returned since recv is called with dontwait
20090518 22:03:49< silene> wild guess: a single thread is responsible for multiple receiving sockets, so it tries to read as much as possible in non-blocking mode, and each time it fails, it skips to the next socket with available data
20090518 22:04:47-!- boucman [n=rosen@wesnoth/developer/boucman] has quit [Read error: 104 (Connection reset by peer)]
20090518 22:05:48< Soliton> there's just one socket.
20090518 22:09:12-!- boucman [n=rosen@wesnoth/developer/boucman] has joined #wesnoth-dev
20090518 22:09:29-!- Elvish_Pillager [n=eli@68-116-175-221.dhcp.oxfr.ma.charter.com] has quit [Read error: 110 (Connection timed out)]
20090518 22:10:10< silene> then i must say i don't see the point of using non-blocking sockets, the sdl functions would be just as good
20090518 22:13:33< CIA-30> ilor * r35738 /trunk/src/network_worker.cpp: remove unused code in send_buffer
20090518 22:13:48-!- giusef [n=giusef@unaffiliated/giusef] has quit ["exit (-1);"]
20090518 22:13:56-!- Polarina [n=polarina@wesnoth/translator/Polarina] has joined #wesnoth-dev
20090518 22:14:06< Soliton> i think that was written when we still had problems with not detecting stale connections. i don't remember if there was a bug with the SDL function or if it was just a desperate try...
20090518 22:14:10< CIA-30> ilor * r35739 /trunk/src/network_worker.cpp: add win32-specific error check to receive_with_timeout instead of ignroing the error
20090518 22:14:16< crimson_penguin> Ivanovic: "msgfmt (GNU gettext-tools) 0.17" - I think I've got it all set up now, so translators could use that feature that'll make things easier (if me not having a new enough msgfmt was the only problem)
20090518 22:14:30< ilor_> well, I added a if-win32 way similar to the EAGAIN
20090518 22:15:02< ilor_> WSAEWOULDBLOCK looks like th equivalent
20090518 22:15:59< ilor_> silene: I removed the dead code (above), I'd prefer to hear your opinion before I commit to 1.6
20090518 22:16:05< boucman> alink: around ?
20090518 22:16:53< Rhonda> Ivanovic: You didn't mention DM in your 1.7.0 announce!
20090518 22:17:27< silene> ilor_: is your previous patch is already committed to 1.6? if so, then please commit, there is no point in keeping the code around; otherwise i would just wait for the trunk version to be exercised
20090518 22:19:02< boucman> Rhonda: yes he did...
20090518 22:21:21< boucman> since 9.04, xorg is a cpu hog here... I can barely compile at all
20090518 22:22:35< loonycyborg> boucman: What kind of videocard you have?
20090518 22:22:50< boucman> Radeon HD 3600
20090518 22:23:06< boucman> I use the default drivers (which are the free ones IIUC)
20090518 22:23:34< Ivanovic> Rhonda: i have
20090518 22:23:46< Soliton> ilor_: you hit the if(true) codepath before? nothing too bad happens though just some unnecessary waiting, right?
20090518 22:23:57< loonycyborg> boucman: You're not using something like compiz-fusion, right? :)
20090518 22:24:03< boucman> nope
20090518 22:24:05< crimson_penguin> make clean isn't working in po/ - that is, after running it, "make" doesn't do anything
20090518 22:24:07< ilor_> Soliton: the socket was bad and it looped
20090518 22:24:09< Ivanovic> Rhonda: in fact even twice
20090518 22:24:14< boucman> disabled with the free drivers, it seems
20090518 22:24:22< Soliton> ilor_: well, and timed out, no?
20090518 22:24:24< Ivanovic> once in "a new campaign was added" and the 2nd time in the more detailed version
20090518 22:24:36< ilor_> not really, and it pegged the cpu
20090518 22:25:00< ilor_> I'm on my laptop and I can hear the fans spinnig up if a core is maxed ;)
20090518 22:27:05< crimson_penguin> it runs "rm -fr *.o" and other things, but nothing to remove the .gmo files
20090518 22:27:05< Soliton> weird, it looks to me as if it should just wait until the timeout and then return false.
20090518 22:27:54< ilor_> the cpu load certainly was there, though I didn't wait long
20090518 22:28:13< ilor_> and at any rate, the win32 specific code looks somewhat proper
20090518 22:28:18< Soliton> less than 30s?
20090518 22:28:21< ilor_> less
20090518 22:28:26< Soliton> ok.
20090518 22:29:09< ilor_> I'll backport both of the above to 1.6
20090518 22:29:12< Ivanovic> hmm, hasn't sirp rewritten some parts of the server to be posix only but reduce memory usage?
20090518 22:29:34< Ivanovic> (as in: problems under windows are to be expected)
20090518 22:29:55< ilor_> well we do ship win32 wesnothd don't we?
20090518 22:30:01< Soliton> yep.
20090518 22:30:04< Ivanovic> at least sirp introduced simplewml
20090518 22:30:20< Soliton> the raw socket part is not compatible with windows i guess but otherwise it should be fine.
20090518 22:30:25< Ivanovic> i am not sure if he rewrote both, campaignd and wesnothd, or only campaignd or wesnothd
20090518 22:30:29< ilor_> simplewml looks portable enough, I just guess the win32 network code wasn't tested much
20090518 22:30:42< Soliton> apparently. :-)
20090518 22:31:22< silene> crimson_penguin: "make maintainer-clean"
20090518 22:31:31-!- Polarina1 [n=polarina@85-220-26-48.dsl.dynamic.simnet.is] has quit [Read error: 104 (Connection reset by peer)]
20090518 22:31:38< CIA-30> ilor * r35740 /branches/1.6/src/network_worker.cpp: remove unused code in send_buffer
20090518 22:31:56< CIA-30> ilor * r35741 /branches/1.6/src/network_worker.cpp: add win32-specific error check to receive_with_timeout instead of ignroing the error
20090518 22:31:58< crimson_penguin> silene: thanks
20090518 22:32:07< silene> crimson_penguin: but be careful that it removes more than just .gmo files
20090518 22:32:19< crimson_penguin> silene: yeah, I have to reconfigure now, but that's OK
20090518 22:32:39< crimson_penguin> I'm just testing my new msgfmt
20090518 22:33:39 * boucman is going to try the ATI non-free drivers
20090518 22:33:42< boucman> wish me luck
20090518 22:34:01< Crab_> good luck :)
20090518 22:34:17< boucman> last time I had to resintall my distro :P
20090518 22:34:25< Ivanovic> boucman: outsch
20090518 22:34:28< Ivanovic> how have you installed it?
20090518 22:34:41< boucman> using the ubuntu non-free driver app
20090518 22:34:43< Ivanovic> normally you should be able to install it via your package manager and thus easily be able to remove it, too
20090518 22:34:55< boucman> Ivanovic: if I can reach a console...
20090518 22:35:11< boucman> last time it manage to stuck the computer so much I couldn't reach the console at all
20090518 22:35:14< Soliton> ilor_: i think you can also change the if(true) to if(false) doesn't seem useful to try again in that case.
20090518 22:35:18< Ivanovic> interesting, i never had such a prob with fglrx...
20090518 22:35:18< boucman> (crash the video card, I guess)
20090518 22:35:26< ilor_> hah, reminds me of, iirc, ubuntu 8.04, which by default wouldn't start X if you have a laptop ati card and a widescreen display
20090518 22:35:57< boucman> rebooting now
20090518 22:36:03-!- boucman [n=rosen@wesnoth/developer/boucman] has quit ["Leaving."]
20090518 22:37:02< Rhonda> Ivanovic: Twice? You mean, like in the forum announce?
20090518 22:37:18< Ivanovic> jupp
20090518 22:37:38< Ivanovic> once i say that a new campaign was added and the 2nd is describing that DM was added, coming from the UMC-dev project
20090518 22:38:06< Rhonda> :)
20090518 22:39:03-!- thespaceinvader [n=chatzill@wesnoth/artist/thespaceinvader] has joined #wesnoth-dev
20090518 22:47:44< Ivanovic> off to bed, n8
20090518 22:49:03-!- stikonas [n=and@wesnoth/translator/stikonas] has quit [Read error: 104 (Connection reset by peer)]
20090518 22:49:13-!- stikonas [n=and@ctv-213-164-125-83.vinita.lt] has joined #wesnoth-dev
20090518 22:52:03-!- ShikadiLord [n=ignacio@wesnoth/developer/shadowmaster] has joined #wesnoth-dev
20090518 22:52:15-!- ShikadiLord [n=ignacio@wesnoth/developer/shadowmaster] has quit [Nick collision from services.]
20090518 22:59:08< CIA-30> ilor * r35742 /branches/1.6/src/network_worker.cpp: in receive_with_timeout, assume an error is fatal when no platform-specific checks are available
20090518 22:59:59-!- EdB [n=edb@182.117.88-79.rev.gaoland.net] has joined #wesnoth-dev
20090518 23:01:14< CIA-30> ilor * r35743 /trunk/src/network_worker.cpp: in receive_with_timeout, assume an error is fatal when no platform-specific checks are available
20090518 23:04:37-!- Noyga [n=lame-z@wesnoth/developer/noyga] has quit ["Quitte"]
20090518 23:04:44< Soliton> Ivanovic: changed the wording about testing 1.7.0 a bit to clarify that we do appreciate testers.
20090518 23:04:54-!- Turuk [n=Turuk@wesnoth/forumsith/turuk] has joined #wesnoth-dev
20090518 23:05:00< Soliton> (in the announcment)
20090518 23:06:00-!- Polarina [n=polarina@wesnoth/translator/Polarina] has quit [Success]
20090518 23:09:14< zookeeper> huh. http://www.wesnoth.org/forum/viewtopic.php?f=4&t=25509
20090518 23:09:38< zookeeper> i encountered that 0gold unit too in a rough life, but i of course assumed that it was just a silly broken campaign feature.
20090518 23:10:06< zookeeper> and that was with 1.6.2
20090518 23:10:08< Turuk> Meh
20090518 23:11:55< stikonas> Is it a known bug that units who stand on void squares can not be harmed?
20090518 23:12:43< Soliton> no.
20090518 23:12:56-!- Polarina [n=polarina@wesnoth/translator/Polarina] has joined #wesnoth-dev
20090518 23:13:12< Soliton> but it is known that units without a defense value will get 100% defense.
20090518 23:14:25< stikonas> I played Dark Forecast
20090518 23:15:03< stikonas> and Drake Fighter appeard on the void terrain
20090518 23:15:21< stikonas> and it was not possible to kill it without retreating
20090518 23:15:27-!- silene [n=plouf@wesnoth/developer/silene] has quit [Read error: 113 (No route to host)]
20090518 23:21:05< Soliton> i guess we could also make it default to 0%.
20090518 23:23:28-!- Blueblaze [n=nick@c-98-199-143-139.hsd1.tx.comcast.net] has joined #wesnoth-dev
20090518 23:24:39< Soliton> infact there're a bunch of other cases where it does default to 0% so i'll adapt that.
20090518 23:24:44< Rhonda> Ivanovic: ... and forgot to mention lua :)
20090518 23:26:20< Soliton> * We now depend on lua >= 5.1.4
20090518 23:26:52-!- LFC|Doppp [n=aassdf@c-67-171-96-240.hsd1.pa.comcast.net] has joined #wesnoth-dev
20090518 23:27:31< Rhonda> * A new dependency was introduced: Lua 5.1.
20090518 23:27:33< Rhonda> Oh, right.
20090518 23:27:40-!- LFC|Doppp is now known as Doppp
20090518 23:27:46< alink> rebooting boucman: i am here now
20090518 23:28:22< alink> btw ~1h to reboot is not normal ;-p
20090518 23:29:15< alink> oops I hope that it is not again: <boucman> last time it manage to stuck the computer so much I couldn't reach the console at all
20090518 23:41:06-!- stikonas [n=and@wesnoth/translator/stikonas] has quit [Read error: 110 (Connection timed out)]
20090518 23:42:24-!- Turuk [n=Turuk@wesnoth/forumsith/turuk] has left #wesnoth-dev []
20090518 23:43:25-!- f6p [n=f6p@dynamic-78-8-65-140.ssp.dialog.net.pl] has joined #wesnoth-dev
20090518 23:51:04-!- EdB [n=edb@182.117.88-79.rev.gaoland.net] has quit [Remote closed the connection]
20090518 23:53:41-!- f6p [n=f6p@dynamic-78-8-65-140.ssp.dialog.net.pl] has quit ["Leaving"]
20090518 23:53:44-!- Crab_ [n=Crab_@wesnoth/developer/crab] has left #wesnoth-dev []
20090518 23:58:24-!- zookeeper [n=l@wesnoth/developer/zookeeper] has quit []
--- Log closed Tue May 19 00:00:49 2009