#exult@irc.freenode.net logs for 20 Feb 2010 (GMT)

Archive Today Yesterday Tomorrow
Exult homepage

[00:18:20] <Marzo> There it is: MinGW GCC 4.4.0 crashed in shapewrite.cc
[00:33:22] <Colourless> http://pentagram.svn.sourceforge.net/viewvc/pentagram/pentagram/trunk/pentagram.cpp?revision=2443&view=markup << 'Ages' ago i SDL_Main related redirection stuff in pentagram
[00:42:08] <-- Colourless has left IRC (Ping timeout: 240 seconds)
[00:42:08] --> Colourless` has joined #exult
[00:42:10] --- Colourless` is now known as Colourless
[00:42:10] --- ChanServ gives channel operator status to Colourless
[00:49:10] <Marzo> That works
[00:49:21] <Marzo> I already have a solution I am trying, though
[00:53:45] <Marzo> (or would be testing if GCC stopped miscompiling... sigh)
[03:23:30] <-- Colourless has left IRC (Quit: casts improved invisibility)
[03:26:30] --> Colourless has joined #exult
[03:26:30] --- ChanServ gives channel operator status to Colourless
[04:03:00] <-- julien has left IRC (Ping timeout: 256 seconds)
[04:03:12] --> julien- has joined #exult
[04:18:56] <Marzo> Argh, enough. MinGW's GCC 4.4.0 is pure crap. Downgrading to another version...
[04:37:38] --> Malignant_Manor has joined #exult
[04:41:21] <Malignant_Manor> Win95 disc is OEM so I have to downgrade Win98 to test. I guess I could format my 428.1 MB hard drive and hope it has enough space.
[04:46:04] <Colourless> 428mb should be more than enough for 95
[04:46:43] <Colourless> just don't start installing anything, except maybe directx
[04:46:53] <Colourless> remember that 95 didn't come with that :-)
[04:47:17] <Malignant_Manor> OEM copy with crap, though. CD is 501.
[04:48:09] <Malignant_Manor> 510 MB
[04:48:30] <Colourless> according to a quick search, install size is 275mb
[04:48:56] <Malignant_Manor> wikipedia says 120 min
[04:49:20] <Malignant_Manor> Oh, well. See what is on disc besides win 3.1
[04:49:42] <Malignant_Manor> Maybe it does have Win95, but I doubt it.
[04:49:57] <Malignant_Manor> I think this was from the Pentium 90.
[04:50:47] <Colourless> might depend on which 95 OSR it is
[04:52:58] <Colourless> the later versions included versions of IE. the original did not
[05:16:53] <Marzo> The basic code for redirecting things to local appdata\Exult is now working; technically, it was working several hours ago, or would have been if GCC 4.4.0 from MinGW wasn't miscompiling the code and causing all manners of segfaults
[05:17:38] <Marzo> I will set ES and UCXT to do it too, and fix screenshots to be saved in the same location, before I commit, though
[05:27:08] <Malignant_Manor> It complains about disc space and quits.
[05:28:59] <Malignant_Manor> Downgrading Win98 should be fine, but upgrading to Windows 98SE had issues with the isa that sometimes made me have to downgrade and reinstall several times.
[05:37:04] <Marzo> Well, good night to all
[05:41:57] <-- Marzo has left IRC (Ping timeout: 260 seconds)
[07:35:20] <Malignant_Manor> I've got an Exult snapshot from the Jan running in Win 95. It seems to run fine except not exiting properly. (need to kill it from task manager))
[07:35:43] <Malignant_Manor> It needs winsocks 2.0.
[07:36:42] <Malignant_Manor> I don't have Exult Studio on it to test though.
[07:37:37] <-- Rottingbeef has left IRC (Ping timeout: 264 seconds)
[07:41:23] --> Rottingbeef has joined #exult
[09:07:18] <Malignant_Manor> kernel32.dll in Windows 95 is missing isDebuggerPresent and won't start.
[09:07:35] <Malignant_Manor> for Exult Studio
[11:38:12] <Kirben> Looks like GTK+ dropped Windows 95 support at some point.
[11:57:20] <Malignant_Manor> It's at least one dll.
[11:58:17] <Malignant_Manor> I think it was libglib-2.0-0.dll.
[12:15:00] <-- Colourless has left IRC (Quit: casts improved invisibility)
[12:55:26] <Malignant_Manor> Current GTK+ only supports Win2k and higher.
[12:59:40] <Kirben> I'm still using older version, which was meant to still support Windows 9x/ME.
[13:01:29] <Kirben> Glib 2.6.6, GTK+ 2.6.9.
[13:01:53] <wjp> is there still any demand for Win95/98/ME compatibility?
[13:04:22] <Malignant_Manor> There are a few that use Win98.
[13:06:32] <Malignant_Manor> If wanting to drop Win98 support, it would be nice to actually have an Exult Studio release first along with Exult.
[13:07:38] <Malignant_Manor> Marzo didn't want to have an ES release atm though.
[13:09:37] <Malignant_Manor> There's been no complaints about ES not running on Win95, so it should be fine to not support it there.
[13:13:01] <Malignant_Manor> At least, supply old build libraries if official support drops a version of Windows.
[13:17:21] <Malignant_Manor> I mean, provide a download and say official support stopped at revision xxxx.
[13:32:00] <-- Kirben has left IRC ()
[14:03:39] <-- Malignant_Manor has left IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6/20100115144158])
[14:47:22] --> Marzo has joined #exult
[16:18:57] --> Dominus has joined #exult
[16:18:58] --- ChanServ gives channel operator status to Dominus
[16:22:24] <Dominus> @Malignant (it seems you read the logs): did you compile Exult & ES by using the compile packages offered by Kirben, or did you just use the current Snapshots?
[16:25:29] <Dominus> IF the compile packages (and thus the snapshots) provided by Kirben are still working on Windows 9x, we could still keep on supporting Windows 9x.
[16:26:09] <Dominus> If not I'm really not sure whether we should keep supporting it, at least for Exult Studio...
[16:26:16] <Marzo> To Colourless, if you read the logs before I point them to you: As it turns out, I think I liked the way Pentagram deals with the SDL stdout/stderr redirection slightly better than what I did and so I am going with it
[16:44:54] --> Malignant_Manor has joined #exult
[16:46:50] <Malignant_Manor> I flip through the logs only once in a great while unless I figure there is discussion.
[16:47:23] <Malignant_Manor> I used Kirben's snapshots. Exult Studio will not work in Win95.
[16:48:06] <Dominus> ok, thanks for that...
[16:48:22] <Dominus> soooooā€¦ maybe bury Windows 9x support...
[16:49:10] <Malignant_Manor> Exult, itself, works in Win95.
[16:49:52] <Malignant_Manor> Exult Studio still works in Win98.
[16:50:00] <Dominus> ahhh
[16:50:24] <Dominus> so it's only not working in lock stock windows 95
[16:51:13] <Dominus> and probably just for lock stock Windows 95, probably for one with all availlable updates it *could* be fine as well?
[16:51:56] <Malignant_Manor> The only reason I know of that Win95 doesn't work is because of at least one GTK+ dll requiring kernel32.dll to have the function 'isDebuggerPresent'.
[16:52:51] <Malignant_Manor> I'm not sure if any specific code requires a higher version of Windows.
[16:54:34] <Dominus> was that with a fully loaded Win95 or was it just fresh install + some updates like DirectX?
[16:54:53] <Malignant_Manor> This could mean having to patch a bunch of files from another project to not use that function (unless it is MSCV adding it).
[16:55:54] <Malignant_Manor> That's any version of Win95 unless there is a solution in a homemade kernel.
[16:59:20] <Malignant_Manor> GTK+ doesn't support anything below Win2k anymore.
[17:01:29] <Malignant_Manor> I have no idea where the cutoff for building any of the libraries would be either when trying to add Win95 support.
[17:04:37] <Malignant_Manor> I don't think there is any home brew kernel update like there is for 98SE/ME.
[17:21:05] <Dominus> but there were updates for w95 that also updated kernel32.dll, weren't there? These don't make the situation better?
[17:24:58] <Malignant_Manor> 'isDebuggerPresent' was never added to the kernel.
[17:25:43] <Dominus> ok, believe you now :)
[17:27:05] <Dominus> I'm taking a note for a ES docs update that Win95 is not supported, but Win98/SE/ME are
[17:27:41] <Dominus> At least as long as we don't add any gtk functionality that the old gtk version that kirben uses supports :)
[17:28:04] <Malignant_Manor> Win95 users also need to install Winsock 2. (likely already installed if using ethernet)
[17:28:30] <Malignant_Manor> That's for Exult Studio. (bit late to comment)
[17:28:59] <Dominus> Huh?
[17:29:55] <Dominus> Because of Exult's windows9x support, users running exult on Windows 95 need winsock 2. <- that's what you mean?
[17:33:34] <Malignant_Manor> Exult Studio would require it, if compatibility was fixed.
[17:34:04] <Malignant_Manor> Not that it really matter since I doubt it is going to get supported.
[17:35:29] <Dominus> ok, so Exult doesn't need it? That's good. I really doubt that it is something we *need* to fix. Though Kirben could maybe fix it for his package...
[17:37:04] <Dominus> got to go again :)
[17:37:08] <Malignant_Manor> I don't even know if he has Windows 95 to test.
[17:37:08] <-- Dominus has left IRC (Quit: Leaving.)
[18:14:14] <-- Malignant_Manor has left IRC (Quit: ChatZilla 0.9.86 [Iceweasel 3.0.12/2009082121])
[19:54:37] --> Cahaan has joined #exult
[21:32:00] <-- Cahaan has left IRC ()
[22:24:02] --> Kirben has joined #exult
[22:24:02] --- ChanServ gives channel operator status to Kirben
[23:18:31] <Marzo> Kirben: FYI, I eventually downgraded from MinGW's GCC 4.4.0 to TDM's 4.3.3-sjlj (http://www.tdragon.net/recentgcc/)
[23:18:54] <Marzo> The 4.4.0 was crashing on ES and miscompiling on Exult
[23:19:11] <Marzo> I recommend staying far away from
[23:19:32] <Marzo> (I haven't tried TDM's 4.4.1 yet, though)
[23:20:29] <Kirben> I still to official MinGW builds of GCC (at least those that work), TDM's ports aren't official or really supported.
[23:20:40] <Marzo> Yes, I know
[23:20:44] <Kirben> Getting two compile errors with current Exult SVN:
[23:20:45] <Kirben> ./files/utils.cc:622: error: continue statement not within a loop
[23:20:45] <Kirben> ./files/utils.cc:629: error: continue statement not within a loop
[23:21:31] <Marzo> I knew I should have testes that in my Windows box
[23:21:51] <Marzo> Originally, I had a loop for CSIDL_* values
[23:22:19] <Marzo> Change lines 622 and 629 to: return string("");
[23:23:47] <Marzo> (the loop went with CSIDL_LOCAL_APPDATA first, then with CSIDL_APPDATA; but any Windows that doesn't support CSIDL_LOCAL_APPDATA from shell32.dll requires another dll for SHGetFolderPath)
[23:24:15] <Marzo> (they are all pre-WinME versions)
[23:25:23] <Marzo> I just committed that fix, by the way
[23:29:22] <Kirben> I'm getting error about path not been set, even though they are set in exult.cfg, when starting Exult now.
[23:31:56] <Kirben> When using Process Monitor that paths look incorrect:
[23:31:58] <Kirben> C:\Games\Exult\(ž"\Exult\exult.cfg
[23:42:19] <Marzo> See if there is an "Exult" folder in "Documents and Settings\(username)\Local Settings\Application Data" and if there is anything in it
[23:42:44] <Marzo> Also, can you post your exult.cfg in pastebin just in case?
[23:43:23] <Marzo> Query: did you compile with debug symbols?
[23:46:57] <Kirben> No, there is no Exult sub directory, at that location.
[23:47:01] <Kirben> Using default Makefile.mingw, DEBUG_SYMBOLS isn't deifned.
[23:47:21] <Marzo> Yes, I was afraid of that
[23:47:40] <Marzo> You said that you are using GCC 4.2-dw2 right?
[23:47:50] <Kirben> http://members.optusnet.com.au/wormmon/exult.cfg
[23:48:09] <Kirben> gcc version 4.2.1-dw2 (mingw32-2)
[23:48:15] <Marzo> Hm
[23:50:12] <Marzo> Since we will need to debug this somehow (as it is working in my end), I suggest you make this modification and recompile: in utils.cc, lines 658 and 696, replace them by "string folderPath;"
[23:50:27] <Marzo> This will put stdout/stderr back in the current dir
[23:55:38] <Kirben> Trying, linker is very slow.
[23:57:52] <Marzo> My entire experience with the more recent MinGW tools has been slow
[23:58:31] <Marzo> I can compile Exult from a clean tree in Ubuntu in the time it takes MinGW to link expack