#exult@irc.freenode.net logs for 11 Oct 2002 (GMT)

Archive Today Yesterday Tomorrow
Exult homepage


[00:24:26] <-- Fingolfin has left IRC ("42")
[01:36:46] --> Kirben has joined #exult
[01:36:47] --- ChanServ gives channel operator status to Kirben
[02:31:33] <-- Dark-Star has left IRC ()
[05:11:50] --> hew has joined #exult
[05:12:00] <hew> Anyone around?
[05:14:59] <Kirben> yes
[05:19:10] <hew> Hey..
[05:20:21] <hew> I have both Serpents and Black Gate I need a way to modify the config file so it will read both and allow them to be in seperate directories, I noticed in the documentation that you have how to do it for windows.. what about OS X?
[05:21:08] <hew> Great job on Exult by the way :D
[05:21:18] <hew> My fave game ever on OS X :P
[05:21:23] <hew> Never though I'd see the day
[05:23:21] <Kirben> Just alter the path used in exult.cfg, see readme.html for details. It is the same way for all platforms.
[05:23:32] <hew> k
[05:23:47] <hew> Thanks..
[07:04:25] --> wjp has joined #exult
[07:04:25] --- ChanServ gives channel operator status to wjp
[07:04:29] <wjp> hi
[07:06:17] --- Darke|afk is now known as Darke
[07:12:38] <hew> Hey
[07:15:16] <Darke> Hi!
[07:15:35] <hew> Hey Darke
[07:16:02] <hew> I still can't get the OS X version to link to my specified directories for Black Gate and Serpent's Isle
[07:17:02] <wjp> did you read the section on exult.cfg in the docs?
[07:17:07] <hew> Yea
[07:17:15] <wjp> twice? ;-)
[07:17:21] <hew> Like 20 times :D
[07:17:39] <wjp> are you familiar with xml-like file formats?
[07:17:56] <hew> Not really
[07:18:22] <hew> I messed around with the config file, had no problem doing that, but it didn't work no matter what I tried
[07:18:55] <wjp> do you have a 'blackgate' section and a 'serpentisle' section
[07:18:56] <wjp> brb
[07:19:12] <hew> Yea
[07:19:16] <hew> I have both
[07:22:17] <hew> I have to go for a bit
[07:22:21] <hew> I'll be back later
[07:22:44] <hew> If you want leave some messages about how to modify the config file for OS X
[07:22:53] <hew> The directory I used should be fine
[07:23:12] --- hew is now known as hew[Out]
[07:47:03] <-- Darke has left IRC ("BRB")
[08:00:24] --> Darke has joined #exult
[08:00:24] --- ChanServ gives channel operator status to Darke
[08:03:36] * Darke hmms... he hasn't looked at the forum in the last week, yet there's been comments in only half a dozen threads. It really must be quiet.
[08:09:41] <-- Darke has left IRC (""BRB... Again"")
[08:11:42] --> Darke has joined #exult
[08:11:42] --- ChanServ gives channel operator status to Darke
[08:13:41] <-- hew[Out] has left IRC ()
[08:52:18] --> Dominus has joined #exult
[08:52:55] --- ChanServ gives channel operator status to Dominus
[08:53:03] <Dominus> hi
[08:54:49] <Dominus> So Kirben, now I can "safely" use gcc 3.2? (just reading the logs and cvs messages)
[09:06:11] <Dominus> now let's see what readme.win32 has to say about what I need :-)
[09:09:21] * Darke pawwaves. Hi.
[09:11:21] <Dominus> hey ho Darke
[09:24:39] * Dominus considers posting big signs in the Docs and FAQ: "This text is for all OS unless stated otherwise!"
[09:25:54] <-- wjp has left IRC (brunner.openprojects.net irc.openprojects.net)
[09:25:58] --> wjp has joined #exult
[09:27:58] <Kirben> Dominus: yes
[09:28:18] <Dominus> good work !
[09:28:26] <Dominus> to both of you!
[09:35:16] <Dominus> grrr! msys installation is crap - they should have a "self-install" option...
[09:37:52] <Dominus> now let's see how it compiles
[09:58:15] <Kirben> Are you using sdl 1.2.5 now too ?
[09:58:25] <Dominus> yep
[10:00:06] <Kirben> Good, just uploading a few tools updates now too. Small libpng update and others recompiled with gcc 3.2.
[10:00:33] <Dominus> ah nice (now that I spent ages compiling exult :-))
[10:00:53] <Kirben> Doubt they will make any difference but I hear its best not to link to libs built with older gcc.
[10:01:00] <Dominus> studio doesn't compile right now because of the execbox (or whatever that is called), right?
[10:01:17] <Kirben> right
[10:01:41] <Kirben> Current exult snapshots were built with latest tools if it is easiesr just to download.
[10:01:50] <Dominus> now on to compiling pentagram viewers
[10:02:34] <Dominus> nah, I want to be able to compile them myself as far as it is possible and after all it doesn't take tooooo long to compile
[10:03:10] <Kirben> exult compile time is quite long, at least on this system.
[10:03:50] <Kirben> ok both dev package updates have been uploaded if you want to check them out.
[10:03:52] <Dominus> it is taking long but still ok (used to be a lot shorter, I think)
[10:04:04] <Dominus> will check out later
[10:04:19] <Dominus> about pentgaram, does it now use sdl_mixer?
[10:04:40] <Kirben> yes, last cvs update seems to make use of it.
[10:04:59] <Dominus> maybe this fixes my bs
[10:05:05] <Dominus> but I doubt it
[10:05:11] <Kirben> what sound card ?
[10:05:26] <Dominus> SB Shite!
[10:05:26] <wjp> hi Kirben, Dominus
[10:05:35] <Dominus> eh, SB Live!
[10:05:38] <Dominus> :-)
[10:05:41] <Dominus> hi wjp
[10:06:05] <Kirben> Using 253 unified drivers ?
[10:06:10] <Kirben> Hi wjp
[10:06:14] <Dominus> yes
[10:07:10] <Dominus> only I have the 252 as the 253 refused to work on my system (original SB Live! Platinum)
[10:07:32] <Kirben> 253 should work as long as you extract drivers manually
[10:07:48] <Kirben> and run CTZAPXX.exe
[10:07:54] <Dominus> these unified drivers seem to be better overall but also seem to trigger a bug with Ryan's midi code
[10:08:19] <Kirben> I don't have my sb live installed right now, I found the drivers are still too noisy at times.
[10:08:37] <Dominus> or better worded: ryan's midi code seems to trigger a bug in those drivers :-)
[10:09:22] <Dominus> uwadv also would bs me from time to time
[10:09:41] <Dominus> and it is using ryans code
[10:10:53] <Kirben> Strange, creative's drivers are always fairly bad but never had them blue screen on me.
[10:12:09] <Dominus> a first for me as well (at least the first time I could blame it so directly at the drivers)
[10:14:23] <Dominus> but with setting the midi device directly to the micrsoft general midi device I don't have crashes in uwadv anymore...
[10:14:33] <Dominus> have to try with pentagram now
[10:16:05] <Dominus> yeah that worked
[10:16:50] <Dominus> Oh, I notice that Ryan's code in pentagram has that error with the volume_curve as well
[10:17:00] <Dominus> wasn't also gcc specific
[10:17:04] <Dominus> ?
[10:17:24] <Dominus> wasn't that also gcc triggered, I mean?
[10:17:41] <wjp> which volume_curve error?
[10:17:47] <Dominus> volume_curve = f instead of a number
[10:17:54] <wjp> oh, that
[10:17:58] <Dominus> was fixed with exult some time ago
[10:18:05] <wjp> that's because our printf implementation doesn't support floats
[10:18:06] <Kirben> Was due to using exult own version of snprintf I think.
[10:18:14] <wjp> snprintf, yes
[10:21:18] <Kirben> I will add colourless's fix to pentagram
[10:21:19] <-- Dominus has left IRC (Read error: 104 (Connection reset by peer))
[10:23:40] --> Dominus has joined #exult
[10:24:16] --- ChanServ gives channel operator status to Dominus
[10:24:24] <Dominus> I just had to try the blue screens again :-)
[10:25:25] <Kirben> Does exult blue screen ?
[10:25:42] <Dominus> not afaik
[10:25:52] <Dominus> haven't tried yet
[10:26:50] <Dominus> oops, exult crashed right away...
[10:28:16] <Kirben> try recompiling pentagram now
[10:28:55] <Kirben> Does pentagram use a gamma setting ?
[10:30:26] --> Fingolfin has joined #exult
[10:30:35] * Dominus doesn't know
[10:30:42] <Dominus> Exult doesn't crash, btw
[10:30:52] <Fingolfin> yo
[10:30:54] --- ChanServ gives channel operator status to Fingolfin
[10:31:17] <Fingolfin> so I guess I am still the only one with any problems, rigth? :-)
[10:31:27] <Dominus> I have to get on Colourless' back when he comes back what is so different between pentagram and exult's midi code
[10:31:32] * Fingolfin remembers wjp mentioning some changes by Ryan, he'll look into them
[10:31:38] <Dominus> hi Fingolfin
[10:32:50] --> Nadir has joined #exult
[10:32:50] --- ChanServ gives channel operator status to Nadir
[10:33:28] <Nadir> hi
[10:33:35] * Dominus is about to bs again now
[10:33:39] <Dominus> hi Nadir
[10:33:58] <Nadir> Darke: early april's fool planning =
[10:33:59] <Nadir> ?
[10:34:43] <Dominus> wjp, Kirben: could it be that the sprintf thing could have caused my bluescreens?
[10:35:00] <Dominus> because I don't get them anymore
[10:35:51] <Fingolfin> hi Nadir
[10:36:02] <Fingolfin> Dominus: "bs" ?
[10:36:07] <Darke> Hi!
[10:36:09] <Dominus> blue screens
[10:36:12] <Fingolfin> ah ok
[10:36:19] * Darke looks innocent. He's not planning for April Fool's. Nope.
[10:36:40] <Fingolfin> hm... I have that funny idea for an April fools
[10:36:50] * Fingolfin wonders how an OS X user would react to a bs, though
[10:36:53] <Dominus> shh
[10:36:58] <Kirben> Dominus: yes it could be the cause I think
[10:37:16] <Dominus> you are not supposed to say anyhting about that here
[10:37:29] * Dominus meant Fingolfin
[10:37:48] * Fingolfin has no clue what Dominus is talking about
[10:38:23] <Dominus> Kirben: could you tell me what the fix was so I can tell vividos of uwadv of it and if it applies there as well
[10:38:36] <Kirben> http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/exult/exult/audio/xmidi.cc.diff?r1=1.58&r2=1.58.4.1
[10:38:48] <Kirben> that is the fix
[10:38:49] <Dominus> ok, thanks
[10:39:38] * Fingolfin notes that exult must have been working for him on 2002-07-27
[10:40:28] <Dominus> which is ages ago
[10:44:47] * Fingolfin would wish he knew what is causing his exult crashes
[10:45:26] <Fingolfin> well, strictly spoken, I know what is causing them ... malloc(4GB). But WTF is it doing that on my system anyway, and not on others? And why only with the BG flx file, and in Setup ?!?
[10:45:43] <Dominus> is it just the main branch or the 1.0.x branch as well?
[10:47:21] <Fingolfin> this should be head I think
[10:47:23] <Fingolfin> HEAD CVS even
[10:47:43] <Fingolfin> yup
[10:47:54] <Fingolfin> which reminds me, I probably should apply my recent changes to the 1.0 branch as well
[10:49:57] <Dominus> yup
[11:01:02] <Nadir> what does gdb say ? (or whatever tool you use in OSX)
[11:01:50] <Nadir> the stacktrace to that malloc
[11:02:03] * Fingolfin points Nadir to the channel logs from last night for extensive back traces :-)
[11:02:42] * Nadir goes to the logs
[11:03:26] * Fingolfin notes that all this was with HEAD CVS, he is right now preparing the 1.0 branch to compile again on his system, after which he'll compile it, and then test... might be longer than I'd like it to be till it's ready, though
[11:04:03] <Fingolfin> is anybody here still using automake < 1.6 or autoconf < 2.52 ? Just curious
[11:04:15] <Fingolfin> (for those of us using them at all that is :-)
[11:05:06] * Darke uses both make/conf 1.6/2.52 and a much earlier version. Why? *grin*
[11:08:57] * Nadir uses Debian, which defaults to automake 1.4, but I have also installed the 1.6 package, which has to be invoked as automake-1.6 :(
[11:10:54] * Fingolfin has installed automake 1.4, 1.5 and 1.6 in parallel
[11:11:12] <Fingolfin> they released a maintanance update of 1.4 which installs it as automake-1.4 very useful, evolution makes use of it
[11:15:44] <Nadir> Fingolfin: your problem looks tricky
[11:18:10] <Nadir> have you guys seen this: http://h-world.simugraph.com/index.html
[11:18:39] <Nadir> by the same guy doing simutrans http://www.s-line.de/homepages/simutrans/simutrans/index.html
[11:19:57] <Dominus> looks nice. a bit like Jeff Vogel'S style
[11:20:56] <Nadir> who is jeff vogel ?
[11:21:10] <Dominus> Spiderweb software
[11:21:14] <Nadir> ah
[11:21:15] <Nadir> exile
[11:21:20] <Dominus> exactly
[11:21:26] <Nadir> simutrans is closed source :(
[11:21:45] * Darke hmms... not too bad. Looks a bit... umm... 'blocky' though.
[11:22:43] <Nadir> http://h-world.simugraph.com/images/garden.png
[11:22:46] <Nadir> this one doesn't
[11:23:20] <Kirben> Anyone want to try replacing getopt in usecode/compiler/ucmain.cc with getopts from http://www.dragon-ware.com/~steve/projects/getopts/ ??
[11:25:37] <Nadir> why?
[11:25:37] <Nadir> :)
[11:25:57] <Kirben> since getopt is no longer available for mingw.
[11:26:11] <Kirben> ucc compile broken since june
[11:26:46] <-- wjp has left IRC (brunner.openprojects.net irc.openprojects.net)
[11:26:46] <-- Dominus has left IRC (brunner.openprojects.net irc.openprojects.net)
[11:26:46] <-- Fingolfin has left IRC (brunner.openprojects.net irc.openprojects.net)
[11:26:46] <-- cnj has left IRC (brunner.openprojects.net irc.openprojects.net)
[11:26:46] <-- Darke has left IRC (brunner.openprojects.net irc.openprojects.net)
[11:27:09] --> Fingolfin has joined #exult
[11:27:09] --> Dominus has joined #exult
[11:27:09] --> wjp has joined #exult
[11:27:09] --> Darke has joined #exult
[11:27:09] --> cnj has joined #exult
[11:28:26] <Kirben> According to mingw developers, getopt.h was only bundled with old mingw version by mistake.
[11:31:43] <Dominus> Kirben: not related to your question but could you take a look at http://fernando.darocha.free.fr/sblive-audigy.htm ? The only link I can find on this page is down at the bottom the official links....
[11:33:45] <Kirben> Dominus: should be two link buttons too, COMPLET and DRIVERS
[11:34:36] <Dominus> probably gets filtered out by my stupid norton firewall....
[11:35:35] <Kirben> buttons seem to be using flash
[11:36:03] <Dominus> yeah, flash is sometimes problematic on the norton thingy
[11:40:39] --> SB-X has joined #exult
[11:40:48] <SB-X> yo
[11:41:02] <Dominus> hi
[11:42:41] <Darke> Hi.
[11:50:17] * Fingolfin notices PGP 8 beta is out
[12:03:53] <-- Kirben has left IRC ("System Meltdown")
[12:32:27] --> Kirben has joined #exult
[12:32:27] --- ChanServ gives channel operator status to Kirben
[12:34:36] <Fingolfin> ouch, 1.0-dev also crashes
[12:34:40] <Fingolfin> weirdo much
[12:34:48] * Darke yawns and curls up into a ball. Time to sleep. Night all!
[12:34:53] <Fingolfin> night Darke
[12:34:56] <Darke> Ick. Goodluck!
[12:35:07] --- Darke is now known as Darke|zzZ
[12:35:40] * Fingolfin notes that since his last test, he updated to a new major OS version (10.1 -> 10.2), new SDL version (1.2.4 -> 1.2.5) and probably also some other things changed
[12:37:58] <Darke|zzZ> Fingolfin: Why not grab the cvs version a few days before your last 'known good' test, and try to compile that? That way you can at least partially narrow it down to either an exult problem, or an OS/sdl problem.
[12:38:35] <Fingolfin> yup that's my plan
[12:45:56] <-- wjp has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:45:56] <-- Dominus has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:46:07] --> Dominus has joined #exult
[12:46:07] --> wjp has joined #exult
[12:47:06] <-- Darke|zzZ has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:47:09] --> Darke|zzZ has joined #exult
[12:47:47] <-- Fingolfin has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:47:47] <-- cnj has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:48:48] --> Fingolfin has joined #exult
[12:48:48] --> cnj has joined #exult
[12:54:41] <-- Fingolfin has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:54:41] <-- cnj has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:54:41] <-- Darke|zzZ has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:54:41] <-- Dominus has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:54:41] <-- wjp has left IRC (brunner.openprojects.net irc.openprojects.net)
[12:54:55] --> Fingolfin has joined #exult
[12:54:55] --> cnj has joined #exult
[12:54:55] --> Darke|zzZ has joined #exult
[12:54:55] --> Dominus has joined #exult
[12:54:55] --> wjp has joined #exult
[13:12:20] <-- Fingolfin has left IRC ("42")
[13:12:20] <-- Nadir has left IRC (Read error: 104 (Connection reset by peer))
[13:22:50] <Dominus> it seems that NWN suffers from the same problem U9 suffered. Forced release by the publisher and patches that never end... At least it seems they don't want to give up on it...
[13:40:09] --> Nadir has joined #exult
[13:40:09] --- ChanServ gives channel operator status to Nadir
[13:40:30] <Nadir> b
[13:40:55] <Dominus> wb
[13:46:14] <Dominus> gotta go, see you later
[13:46:21] <-- Dominus has left IRC ("enough for now")
[13:56:53] <-- Nadir has left IRC ("Uscita dal client")
[14:48:08] <-- wjp has left IRC ("gtg")
[15:20:28] --> hew has joined #exult
[15:20:46] <hew> I'm back, anyone here?
[15:21:40] <SB-X> yeah
[15:21:55] <SB-X> but the devs left
[15:22:47] <hew> Ahh.. you wouldn't have any idea how to modify the exult.cfg in OS X to set my own directory location for Black Gate and Serpents Isle?
[15:26:12] <SB-X> same as all other systems
[15:27:02] <hew> If it was then I wouldn't be having this problem :D
[15:27:12] <hew> The config file looks completely different
[15:27:29] <hew> Oh wait!
[15:27:34] <hew> I know!
[15:28:28] <SB-X> there is only one format for the config file
[15:29:11] <hew> Got it
[15:29:21] <hew> I was missing the Static path in there :D
[15:29:30] <hew> Couldn't find the gamefiles :P
[15:30:31] <SB-X> eh
[15:30:33] <SB-X> that works?
[15:31:03] <hew> Well I was missing the static tag so it couldn't load the files from my two ultima 7 folders
[15:31:48] <-- Kirben has left IRC ("System Meltdown")
[15:38:14] <SB-X> are your ultima 7 folders called static?
[15:39:07] <SB-X> it should pick that up automatically
[15:39:30] <SB-X> my path is set as /home/sbx/ultima7/bg
[15:56:03] <hew> Mine are setup
[15:56:14] <hew> They're called blackgate and serpentisle
[15:56:18] <hew> Gotta go
[15:56:25] <-- hew has left IRC ()
[15:57:03] <SB-X> htmm
[15:57:05] <SB-X> grmm
[15:57:07] <SB-X> gasd
[15:57:12] <SB-X> hrmm!
[15:57:16] <SB-X> there it goes
[15:57:21] * SB-X sighs.
[15:57:24] <-- SB-X has left IRC ("X-Chat [1.6.4]")
[17:06:54] --> wjp has joined #exult
[17:06:54] --- ChanServ gives channel operator status to wjp
[17:39:14] --> Fingolfin has joined #exult
[17:40:53] <Fingolfin> hiya
[17:40:57] <wjp> hi
[17:44:51] <wjp> did you figure out why exult was crashing, btw?
[17:45:07] <Fingolfin> not yet, I just built an older CVS snapshot, let me test if that crashes
[17:45:15] <wjp> how old?
[17:45:22] <Fingolfin> BTW why does exult.cfg have these <title> elements?!?
[17:45:44] <wjp> well, I think they'll be used for custom games
[17:45:45] <Fingolfin> crash in the old snapshot, too! so it's most probably something that is caused by 10.2
[17:46:34] <Fingolfin> or rather, caused by the interaction between exult and 10.2, not necessarily an OS bug
[17:48:31] <wjp> or it could be an STL bug
[17:50:04] <Fingolfin> possible, yeah
[17:50:33] <Fingolfin> hey, I should try my other (MacOS classic) build of Exult I guess
[17:50:43] <wjp> I'd probably try stepping through the failing Shape::read
[17:52:48] <Fingolfin> yup yup
[17:53:09] <Fingolfin> it's a bit annoying to use GDB from the CLI all the time, though
[18:12:02] --> Dark-Star has joined #exult
[18:20:50] <wjp> hi
[18:25:19] * Fingolfin is in the call to Read4 that returns (uint)-1
[18:25:21] <Fingolfin> _M_streambuf_state = 6,
[18:26:40] <Fingolfin> // 27.4.2.1.3 Type ios_base::iostate
[18:26:40] <Fingolfin> static const __int_type _S_badbit = 0x01;
[18:26:40] <Fingolfin> static const __int_type _S_eofbit = 0x02;
[18:26:46] <Fingolfin> so it seems that it think EOF is reached
[18:26:52] <Fingolfin> thus returning -1
[18:26:58] <Fingolfin> static const __int_type _S_failbit = 0x04;
[18:27:00] <Fingolfin> err
[18:27:05] <Fingolfin> so state = fail | eof
[18:28:43] <wjp> hm, I see
[18:29:20] <wjp> is this in old or new cvs, btw?
[18:29:28] <Fingolfin> I am back to HEAD CVS
[18:29:32] <Fingolfin> as I got the same crash in the old checkouts
[18:29:47] <wjp> k, so this is somewhere in streamdatasource?
[18:31:18] <Fingolfin> Shape::read calls read4 twice, on object "shapes", which is a StreamDataSource
[18:31:23] <Fingolfin> (all according to gdb=
[18:32:27] --- Dark-Star is now known as Dark-Star|afk
[18:33:48] <wjp> right, so the question is where this eof bit is set
[18:34:23] <wjp> hm, there's a seek to near the start of the file earlier in Shape::read
[18:35:25] <Fingolfin> you mean the shapes->seek(shapeoff);
[18:35:28] <Fingolfin> yeah
[18:35:51] <wjp> is it already eof before that?
[18:36:10] <Fingolfin> shapenum=23 -> shapeoff = 0x80 + shapenum*8 = 312
[18:36:15] <Fingolfin> good question
[18:36:20] * Fingolfin changes the code
[18:40:47] <wjp> hmm... according to this part in Josuttis you might need to call clear() on the stream
[18:49:23] <wjp> yes... try adding an in->clear() right before in->seekg(pos) in StreamDataSource::seek()
[18:49:43] * Fingolfin comes back to his machine and find that gcc is still compiling the same file. For the past 10 minutes now. Uhmmm
[18:49:51] <Fingolfin> HD full ouch
[18:49:56] <wjp> ouch
[18:53:32] * Fingolfin removes 200 MB pictures from his system parition - why the heck where those there?!?
[19:30:14] --- Dark-Star|afk is now known as Dark-Star
[19:40:35] <Fingolfin> shapes not good (1)!
[19:40:36] <Fingolfin> shapes->seek(312)
[19:40:36] <Fingolfin> shapes not good (2)!
[19:40:36] <Fingolfin> shapelen (2) = 4294967295
[19:40:44] <Fingolfin> -> the stream seems to be bad *before* the seek already...
[19:40:59] <wjp> yeah, I expected that
[19:41:01] <Fingolfin> (calling the "good()" mehod which returns true if status is 0
[19:41:03] <Fingolfin> yeah
[19:41:19] <wjp> hence the suggestion to clear() the stream before seeking
[19:41:28] <wjp> (which is most likely what we want)
[19:41:43] <wjp> something seems to have read to EOF earlier
[19:42:09] <Fingolfin> yup
[19:42:46] <Fingolfin> so the question is, who seeks over the end?
[19:42:50] <Fingolfin> hmm
[19:42:54] * Fingolfin soundly remembers something
[19:42:59] * Fingolfin checks his email archive
[19:46:06] <wjp> I should really read the section on streams in Stroustrup entirely sometime
[19:46:09] <Fingolfin> http://www.mail-archive.com/darwin-gcc3@opensource.apple.com/msg00209.html
[19:46:19] <Fingolfin> -> might be a bug in OS X stl after all
[19:46:29] <Fingolfin> (which should just be the "normal" gnu stdc++ lib)
[19:46:41] <Fingolfin> this would explain our problem, wouldn't it?
[19:47:00] <wjp> hm, yes, sounds very probable
[19:47:10] <wjp> hm, recent email
[19:47:18] <wjp> hm, I use 'hm' too often :-)
[19:47:27] * Fingolfin writes a reply to the ML
[19:47:47] <Fingolfin> I can also submit a bug report with Apple but help will probably take some months, so for now we'd have to work around it (argh!)
[19:48:04] <wjp> argh indeed
[19:48:33] <wjp> can you reproduce the problem he mentions with his test program?
[19:50:40] <wjp> I wonder what exactly the conditions are for it to trigger
[19:51:59] <Fingolfin> yes
[19:52:05] <Fingolfin> same bug over here
[19:52:14] <Fingolfin> what Visual C++ version are our windows users using?
[19:52:19] <wjp> maybe file size? value of last byte?
[19:52:24] <wjp> not a clue
[19:52:27] * Fingolfin is writing an email to darwin-gcc3 list, stating all systems for which the code works right
[19:57:08] <wjp> hm, this readtest works on the osx machine in the SF compile farm, btw
[19:57:20] <wjp> using gcc 2.95.2
[19:58:05] <Fingolfin> yup
[19:58:09] <Fingolfin> which still runs 10.1
[19:58:15] * wjp nods
[19:58:17] <Fingolfin> and I bet it works here when I use 2.95, too
[19:58:19] * Fingolfin tries that
[19:58:34] <Fingolfin> it does
[20:00:04] <Fingolfin> sent, with CC to you
[20:01:08] <wjp> it's kind of weird that SI's mainshp.flx does not trigger the problem
[20:01:40] <wjp> but maybe that's just because the shapes are in a different spot in the file or something
[20:01:53] <wjp> does it skip the last byte if you run readtest on it?
[20:04:33] <wjp> hm, I wonder what version of osx the macosx workstations at our CS department are running
[20:09:03] <wjp> hm, nice, I seem to have an account on them
[20:09:12] <wjp> but no homedirectory?
[20:12:38] <wjp> hm, anyway, they have gcc 2.95.2 installed
[20:21:12] <Fingolfin> sw_vers will tell you the system version
[20:21:26] <Fingolfin> I guess it's just by chance that for BG we read the last shape in the file
[20:24:34] <wjp> ah, 10.1.5
[20:25:11] <wjp> there are two macos rooms, but the second room doesn't accept ssh connections
[20:26:12] <wjp> uh, correction, they're not even turned on it seems
[20:26:46] <Fingolfin> lol
[20:26:56] <Fingolfin> that would explain why they don't allow ssh, wouldn't it? :-)
[20:27:08] <wjp> yes, I have a feeling it would :-)
[20:52:43] <Fingolfin> anyway I see no viable way to work around this problem short of not using ifstream...
[20:53:35] <wjp> alternate ways of reading from it, maybe?
[20:54:20] <wjp> does the ifstream's rdbuf work properly? (not that I really know what I'm talking about here)
[20:56:50] <Fingolfin> I didn't check out what part of the code is actually wrong
[20:57:07] <Fingolfin> and in any case I am surprised to see this bug - after all the C++ lib should be the same on linux and osx
[20:57:25] <Fingolfin> though Apple GCC is not based on final gcc 3.1 - maybe they left out an important bug fix ?!?
[20:57:50] <Fingolfin> maybe I can get myself the stdc++ headers of linux gcc 3.1 and run a diff to see ...
[21:02:54] <-- Dark-Star has left IRC (brunner.openprojects.net irc.openprojects.net)
[21:02:54] <-- cnj has left IRC (brunner.openprojects.net irc.openprojects.net)
[21:03:08] --> Dark-Star has joined #exult
[21:03:08] --> cnj has joined #exult
[21:26:39] * Fingolfin discovers that istream.tcc is indeed modified in crucial places which affect eof
[21:26:40] <Fingolfin> ha
[21:26:49] <Fingolfin> and in 3.1.1, these places were changed, too
[21:27:03] <Fingolfin> evidence suggests that there was a bug here, and Apple engineers fixed it "the wrong way" ?!
[21:28:32] * Fingolfin replaces his istream.tcc with the 3.1.1 one (after making a backup of course)
[21:29:13] * Fingolfin wonders what happend to his mail to darwin-gcc3
[21:40:37] <wjp> I got it
[21:44:13] <Fingolfin> in any case, seems I either can't test this w/o recompiling stdc++ lib, or it just doesn't help - in any case, the mini test program still behaves wrongly
[22:30:55] <wjp> hm, recompiling libstdc++ doesn't sound like much fun
[22:31:11] <wjp> this is really a very annoying bug
[23:04:00] <Fingolfin> it is
[23:10:02] <-- Fingolfin has left IRC ("42")
[23:21:17] --> hew has joined #exult
[23:54:57] <wjp> hi again
[23:55:03] --- Darke|zzZ is now known as Darke
[23:55:07] <Darke> Hi!
[23:55:19] <wjp> morning :-)
[23:56:41] <Darke> wjp: Umm... aren't you meant to get some sleep or something? Or did you catch a couple of hours of shuteye earlier when you disappeared? *grin*
[23:57:19] <wjp> sleep? hm, yes, probably
[23:57:32] <wjp> I was watching tv earlier
[23:59:54] <-- hew has left IRC (Remote closed the connection)