2008/04/28
IRC [01:13] *** hgcphoenix joined the chat.
IRC [02:52] *** hgcphoenix parted the chat.
IRC [12:28] <Dossy> So, who wants to help merge fixes from CVS HEAD into aolserver_v45_bp branch?
IRC [12:37] *** KNReC joined the chat.
IRC [12:40] *** KNRe1 joined the chat.
IRC [12:41] <Dossy> Hi, Rick.
IRC [12:41] <Dossy> I just asked: So, who wants to help merge fixes from CVS HEAD into aolserver_v45_bp branch?
IRC [12:41] <Dossy> :-)
IRC [12:42] <KNRe1> well, I'm no CVS expert, but I suppose I can lend a hand and get some experience doing it.
IRC [12:42] <KNRe1> What's the procedure?
IRC [12:58] *** KNReC parted the chat.
IRC [13:18] *** daguz parted the chat.
IRC [13:31] *** EvilOtto joined the chat.
IRC [14:41] <Dossy> Check out the aolserver_v45_bp branch. Then, cvs diff -rHEAD ...
IRC [14:41] <Dossy> figure out what changes in HEAD should be in the branch, and "make" those changes by applying the relevant parts of the diff using patch :)
IRC [14:42] <Dossy> There's probably better ways to merge, but I'm lazy :)
IRC [14:45] <EvilOtto> "cvs update -jHEAD -jaolserver_v45_bp file..." or something like that.
IRC [14:45] <Dossy> Well, that takes _all_ the changes ...
IRC [14:45] <Dossy> Are we comfortable with everything in CVS HEAD now?
IRC [14:46] <EvilOtto> well I was saying on a file-by-file basis... or you could do it on a rev by rev basis... doing it on a block-by-block basis is tougher (and why people get excited about mergetools in git or perforce or whatever).
IRC [14:46] <EvilOtto> does HEAD pass all the tests :)
IRC [15:07] <Dossy> :)
IRC [15:07] <Dossy> EvilOtto: That's a trick question. :-)
IRC [15:08] <Dossy> I guess HEAD is safe enough to merge completely into the aolserver_v45_bp branch...
IRC [15:09] <EvilOtto> "2000/10/09: Smoke-test yout server with these ADP scripts. More and more tests will be added soon."
IRC [15:10] <Dossy> Yeah, I know, I know ... :)
IRC [15:11] <EvilOtto> glancing over the diffs, the only things that jump out at me are the pools changes (since they take a fair amount of thought to understand how the logic actually works) and the handling of redirects in the config from server.c (since this is the kind of thing that should be handled in tcl code, IMHO).
IRC [15:16] <EvilOtto> oh, and the conn->channel stuff too. But my philosophical disagreements shouldn't prevent the code from moving forward... is there a word "realprogramming" or "realdevelopment" that expresses the same sentiment as "realpolitik"?
IRC [15:39] <Dossy> hah.
IRC [15:40] <Dossy> realroadmap.
IRC [15:40] <Dossy> :)
IRC [15:40] <Dossy> I think we should just merge everything since enough folks have been using CVS HEAD anyway.
IRC [15:44] <EvilOtto> were you able to replicate my nssqlite3 crash?
IRC [16:29] <Dossy> EvilOtto: nope.
IRC [16:29] <Dossy> [28/Apr/2008:16:29:00][15149.2978036656][-nscp:2-] Notice: nscp: dossy 21: set d [ns_db gethandle soapboxdb]
IRC [16:29] <Dossy> [28/Apr/2008:16:29:00][15149.2978036656][-nscp:2-] Notice: dbdrv: opening database 'sqlite3:/home/dossy/www-data/soapbox.db'
IRC [16:29] <Dossy> [28/Apr/2008:16:29:00][15149.2978036656][-nscp:2-] Notice: nscp: dossy 21: done
IRC [16:29] <Dossy> [28/Apr/2008:16:29:05][15149.2978036656][-nscp:2-] Notice: nscp: dossy 22: ns_db releasehandle $d
IRC [16:29] <Dossy> [28/Apr/2008:16:29:05][15149.2978036656][-nscp:2-] Notice: nscp: dossy 22: done
IRC [16:29] <Dossy> Of course, that's w/ SQLite 3.3.8 or whatnot
IRC [16:29] <Dossy> I'll try to build a newer SQLite3 and see.
IRC [16:30] <Dossy> I just came up with a feature request. There's no way to get the current ADP's file path, if it exists.
IRC [16:30] <Dossy> Hm. What to do ...
IRC [16:31] <EvilOtto> the problem was not in sqlite, it's an uninitialized dbhandle->statement (or context) when an uninitialized valus might not be null.
IRC [18:15] *** Mooooooo joined the chat.
IRC [18:20] <Dossy> EvilOtto: Hm - but, how do I reproduce the crash? That's what's puzzling...
IRC [18:24] <EvilOtto> one quick way would be to patch ns_malloc to memset the allocated memory to something like 0xef. should make no difference to working code, but anything that expects malloced memory to in fact be calloced will be unwell.
IRC [18:24] <EvilOtto> or just otherwise using some debugging malloc that does such a thing already.
IRC [18:24] <Dossy> IIRC, AOLserver's allocator already sets memory to a magic string ... ?
IRC [18:25] <Dossy> But, yeah - I'm sure you're right - you want to commit a fix? :)
IRC [18:28] <EvilOtto> sure.
IRC [19:07] <Dossy> Yeah, I figure since you can already repeat the problem, you're in the best position to work on the fix and commit it :)
IRC [19:09] <EvilOtto> I'll restrict my patch to the crash bug and not yet the DML issue I previously brought up
IRC [19:12] <EvilOtto> I don't have sf commit access tho
IRC [19:12] <Dossy> oh - what's your SF username?
IRC [19:12] <EvilOtto> dvrsn
IRC [19:17] <Dossy> Man, I'm addicted to startup websites. This invite-only thing ... limiting supply seems to exaggerate the limited demand. :-)
IRC [19:21] <Dossy> EvilOtto: OK, you should have CVS commit access now.
IRC [19:23] * EvilOtto laughs evilly: Mua-ha-ha-ha-ha!
IRC [19:23] <Dossy> :-)