#pentagram@irc.freenode.net logs for 26 Feb 2006 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage


[00:03:43] <Colourless> hiya
[00:04:00] <Colourless> so we're setup to use svn now?
[00:04:09] <wjp> yeah
[00:04:26] <wjp> I sent a mail to pentagram-devel about it, but it doesn't appear to have arrived yet
[00:04:27] <Colourless> guess i had better check it out then
[00:04:40] <wjp> my ISP might have gotten itself on some blacklist again :/
[00:05:16] <Colourless> my isp got onto sf blacklist somehow last year
[00:06:00] <Colourless> i think it happens because of mass mailing worms on customer machines
[00:06:17] <wjp> yeah, I think so too
[00:06:25] <wjp> http://svn.sourceforge.net/viewcvs.cgi/pentagram/
[00:07:01] <wjp> please ignore the fact that some files appear committed by 'wjp' instead of 'wjpalenstijn' :/
[00:07:02] <Colourless> revision 2146...
[00:08:17] <wjp> 723 revisions in exult's pentagram module followed by 1421 revisions in our own CVS, and then 1 here locally, and 1 on the actual SVN server
[00:09:06] <Colourless> i installed subversion on a machine here after reading the scummvm stuff. The renaming directories feature was a very useful.
[00:09:25] * wjp nods
[00:19:20] <watt> interesting..... subversion is claiming the certificate isn't from a trusted authority???
[00:19:53] <watt> probably something off on my side, I'd imagine
[00:22:15] <Colourless> i've got it too
[00:22:39] <Colourless> it's probably correct though..
[00:22:51] <wjp> hm, I don't remember seeing it, but I might've got it when I first checked out scummvm early this month
[00:23:07] <Colourless> it isn't important for SF to pay the thousands of $$$ to get a trusted certificate when they aren't doing things with money
[00:23:37] <watt> eh. well, svn 1.3.0 is available and I'm at 1.2.3, so I guess I'll update.
[00:23:45] <wjp> sf's website does have a valid certificate signed by Equifax, though
[00:23:49] <Colourless> then again might just be a svn issue
[00:23:56] <Colourless> should be the same one
[00:24:14] <Colourless> Error validating server certificate for 'https://svn.sourceforge.net:443':
[00:24:18] <Colourless> - The certificate is not issued by a trusted authority. Use the
[00:24:18] <Colourless> fingerprint to validate the certificate manually!
[00:24:18] <Colourless> Certificate information:
[00:24:18] <Colourless> - Hostname: *.sourceforge.net
[00:24:18] <Colourless> - Valid: from Dec 8 13:40:07 2005 GMT until Feb 7 13:40:07 2007 GMT
[00:24:18] <Colourless> - Issuer: Equifax Secure Certificate Authority, Equifax, US
[00:24:20] <Colourless> - Fingerprint: 49:b8:cb:87:04:8c:49:39:45:83:dd:4c:cf:c7:54:57:b0:9e:84:5d
[00:24:41] <Colourless> same as the website
[00:25:43] <watt> svn might not like the wildcard.
[00:26:16] <watt> and spit out a generic message.
[00:26:54] <watt> either way.. I'll update and try again later.
[00:26:55] <Colourless> doesn't really matter. just need to press 'p' for permanent
[00:38:34] <wjp> I hope things work out :-)
[00:38:45] <wjp> maybe I'll try to experiment a bit by reorganizing docs/ tomorrow
[00:38:59] <wjp> moving files around will come in useful there
[00:39:31] <wjp> I'm planning to move file format docs to a subdirectory
[00:39:54] <wjp> anyway, I should be going now; bye
[00:41:47] <Colourless> cya
[03:37:52] <-- Fingolfin has left IRC ("42")
[04:39:40] --- LordNAway is now known as Lord_Nightmare
[09:34:09] <Colourless> a suggestion
[09:34:19] <Colourless> in IDataSource instead of this:
[09:34:20] <Colourless> virtual bool eof() { in->get(); bool ret = in->eof(); if (!ret) in->unget(); return ret; }
[09:34:28] <Colourless> perhaps we should use:
[09:34:28] <Colourless> virtual bool eof() { return in->peek() == std::char_traits<char>::eof(); }
[09:35:17] <Colourless> peek gets the next character without advancing the file pointer, or returns EOF if the next char is eof
[10:03:49] <Colourless> well, managed to get pentagram to compile in windows x64. SDL is the only problem with it. The old directx interfaces are not supported any more so it needs to be rewritten
[10:04:30] <Colourless> runs using only WinDIB. IS some extreme latency with the sound though
[10:30:07] <-- Colourless has left IRC ("casts improved invisibility")
[12:29:06] <wjp> Darke2: your subscription to pentagram-devel was disabled because of bounces
[12:50:58] <Darke2> Yeah. If you can, can you manually remove me from the mailing list? I don't have access to my darkepaw sf.net account because I forgot my password and my ISP at the time was blocking all password change emails from sf.net. Of course the reason I found out about my password not working was because I needed to change my email address because I was swapping ISPs.
[12:51:35] <wjp> I probably can
[12:53:02] * Darke2 need to try to get sf.net to sort it out at some point in time, but he can still commit/etc because he uses pgp stuff to do so, which is why he forgot his password. *grin*
[12:53:26] <wjp> done
[12:53:34] <Darke2> Yay! Thanks.
[13:45:12] <-- Kirben has left IRC (Read error: 110 (Connection timed out))
[18:09:57] <watt> Yay! Compiling!
[18:38:14] <watt> oh good. Our svn:ignores are already set ;-)
[18:38:33] <wjp> yeah, cvs2svn took care of that
[18:43:26] --> SB-X has joined #pentagram
[18:54:35] <wjp> it's rather annoying that zlib's deflate produces valgrind errors
[18:55:00] <wjp> it might be masking saving real uninitialized data :/
[19:06:57] <wjp> I wonder if we should add a configure flag for valgrind-specific code
[20:36:26] <wjp> watt: did you ever look at valgrind's valgrind.h and memcheck.h?
[20:39:10] <wjp> it has some functions to mark blocks as newly allocated, which might be nice to integrate with the custom memory allocators
[20:50:16] <-- SB-X has left IRC ()
[21:19:48] --> Kirben has joined #pentagram
[21:19:48] --- ChanServ gives channel operator status to Kirben
[22:24:53] --> Colourless has joined #Pentagram
[22:24:53] --- ChanServ gives channel operator status to Colourless
[22:29:48] <wjp> hi Colourless
[22:30:03] <Colourless> hi
[22:30:26] <wjp> that IDataSource change you mentioned does look cleaner than the current solution
[22:30:50] <wjp> does it work well when something had already tried to read past eof?
[22:33:06] <Colourless> the implementation i've got here returns EOF if there if the stream is not 'ok' so it should return eof if the stream has gone eof
[22:34:43] <Colourless> baiscally it will either return the next character if it can, or return EOF if it can't
[22:35:24] <wjp> ok, sounds good
[22:39:45] <wjp> will you commit it?
[22:42:07] <Colourless> yeah i'll commit it
[22:42:26] <wjp> if you have a recent exult checkout around, exult's databuf uses it too
[22:42:43] <wjp> if not, I'll take care of that
[22:46:28] <wjp> I'm experimenting with valgrind a bit
[22:46:48] <wjp> I think I got it to recognize the SegmentedPool memory allocations
[22:47:16] <wjp> so it'll properly detect out-of-bounds accesses of things allocated in there, even when they're still within the large buffer allocated with the standard new[]
[22:50:04] <Colourless> my exult checkout is old
[22:52:32] <Colourless> anyway, committed it
[22:52:50] <wjp> just got the email
[23:14:13] <-- Lord_Nightmare has left IRC (Read error: 110 (Connection timed out))
[23:40:01] --> Lord_Nightmare has joined #pentagram