| === Hasone_ is now known as Hasone | ||
| tjaalton | ugh, my stomach.. | 06:38 |
|---|---|---|
| bryce | tjaalton: :-) | 07:52 |
| tjaalton | getting better ;) | 07:53 |
| tjaalton | but I had to skip breakfast | 07:53 |
| bryce | better have plenty of water | 07:54 |
| tjaalton | yeah | 07:54 |
| bryce | I think the openweek session went well, lots of good questions | 07:57 |
| bryce | not yet evidence of new contributors, but of course there's lots of time yet | 07:58 |
| tjaalton | I lurked there for a while | 07:59 |
| tjaalton | but was feeling a bit tired, so didn't raise my voice | 08:02 |
| bryce | no prob, hopefully I didn't get anything too wrong | 08:02 |
| tjaalton | I'll check the logs now :) | 08:03 |
| tjaalton | the OneSecondX fixes should be in xserver 1.5 | 08:05 |
| tjaalton | at least some of them | 08:05 |
| === rzr is now known as zRr | ||
| bryce | heya Q-FUNK | 08:33 |
| Q-FUNK | hey | 08:33 |
| Q-FUNK | bryce: what's up? :) | 08:34 |
| bryce | Q-FUNK: at the moment working on prepping an sru for some -intel quirks, and rebuilding my dev box | 08:35 |
| Q-FUNK | ok | 08:35 |
| Q-FUNK | any news on merging the PCI ID patch to add support for "geode" ? | 08:36 |
| bryce | Q-FUNK: hey btw Marc Tardif mentioned that maddog was very pleased with how things turned out with geode | 08:36 |
| bryce | "This is all thanks to the combined efforts | 08:36 |
| bryce | of David Bensimon, Bryce Harrington and Martin-Éric Racine." | 08:36 |
| Q-FUNK | bryce: obviously he was. they made poor Jordan jumpt through hoops to produce a libDDC patch overnight. | 08:36 |
| Q-FUNK | (or well, maddog did, when he pressured AMD's VP) | 08:37 |
| tjaalton | Q-FUNK: the patch is waiting in git | 08:37 |
| Q-FUNK | btw, I finally reached an agreement with Jordan to release that patch as a new upstream, if we need to, to really close this issue. | 08:37 |
| Q-FUNK | (the libDDC patch) | 08:38 |
| bryce | cool | 08:38 |
| Q-FUNK | my question to you both: which one would be more acceptable as an SRU: running autoconf ourselves on a patched 2.8.0 or having upstream do that for us on what we would call 2.9.0 ? | 08:39 |
| bryce | what would running autoconf attain? | 08:39 |
| Q-FUNK | we need to regenerate the symbols to get the build to take into consideration a new geode_ddc.c | 08:40 |
| bryce | ok, what does the new geode_ddc.c do? | 08:41 |
| Q-FUNK | doing this ourselves on a home-patched 2.8.0 produces a pointlessly huge diff, just for the new config.sub & co | 08:41 |
| bryce | right | 08:41 |
| Q-FUNK | it replaces reliance upon BIOS environment to produce EDID with the X.org libddc that performs the same in a BIOS-free way. it's what fixes it for Koolu. | 08:42 |
| bryce | if it can wait until 8.04.1 I think that sounds like it'd give decent justification for 2.9.0; assuming that 2.9.0 is mostly bug fix (and maybe reasonably safe/lightweight feature work) | 08:42 |
| Q-FUNK | we'd only bump the minor because we replaced code segments with a different implementation. basically, 2.9.0 would indeed be applying that libDDC patch that's already attached to the bug, then doing autoreconf | 08:44 |
| Q-FUNK | and that's all | 08:44 |
| Q-FUNK | we'd move the goal to complete OLPC integration to the next upstream | 08:45 |
| Q-FUNK | but then that would be for intrepid | 08:45 |
| bryce | then it sounds like getting 2.9.0 in for 8.04.1 should be pretty straightforward | 08:45 |
| bryce | getting it SRU'd though would probably be a challenge... they seem to prefer individual patches (and small ones), however the fix sounds quite appropriate in scope | 08:45 |
| Q-FUNK | well, either we run autoconf or upstream does. in either case, the code change itself is only that patch. | 08:46 |
| bryce | sru'ing the autoconf changes on the current 2.8.0 -amd would probably be a decent place to start. If the RM's balk at that, offering a packaged 2.9.0 would be a good fallback | 08:47 |
| bryce | worst case, we could probably fish out the exact Makefile.in changes needed to make the thing work. | 08:47 |
| bryce | maybe if we can show clearly that 2.9.0 is strictly just the one patch, the RM's would be amenable to that | 08:48 |
| Q-FUNK | hm... we'd need several Makefile.in, actually | 08:58 |
| Q-FUNK | and then if any moro decides to build from source, he'd have an out of sync Makefile.am and aclocal.m4 | 08:59 |
| bryce | if we have to patch Makefile.in(s) individually, we'd need corresponding Makefile.am changes. I've done this before; it's a lot of work and quite error prone. | 09:04 |
| bryce | imho better to have a huge autoconf patch than take the risk of hand-crafted Makefile.[am|in] fixups, but whatever gets the job done I guess | 09:05 |
| Q-FUNK | I wouldn't take a risk with cherry picking files from an autoreconf. we really need to take the whole change as a lot. | 09:11 |
| Q-FUNK | btw, if you want to compare a pristine 2.8.0 with one that has autoreconf, debian has a +git release | 09:11 |
| Q-FUNK | I used the libtool, autoconf and automake from Hardy to produce it | 09:13 |
| Q-FUNK | added as comment to the bug | 09:43 |
| Q-FUNK | bryce: btw, where did this message by Marc pop out? | 09:44 |
| Q-FUNK | tjaalton: is the comming for the geode PCI ID made as "geode" or "amd" ? | 09:45 |
| Q-FUNK | tjaalton: upstream and I would prefer "geode" | 09:46 |
| bryce | heya james_w | 09:49 |
| bryce | Q-FUNK: was a private email | 09:49 |
| james_w | hi bryce | 09:49 |
| james_w | how are you? | 09:49 |
| bryce | james_w: doing good. you? | 09:49 |
| james_w | good thanks | 09:49 |
| tjaalton | Q-FUNK: geode | 09:52 |
| Q-FUNK | bryce: ok. it's at least good to know that my efforts are appreciated. | 10:01 |
| bryce | Q-FUNK: I'll forward you the entire email | 10:06 |
| Q-FUNK | ah, that would be nice. thanks! :) | 10:07 |
| bryce | sent | 10:17 |
| Q-FUNK | pretty nice email indeed | 10:17 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!