May 14 18:00:45 * ferringb|afk (~ferringb@ferringb.developer.gentoo) has joined #gentoo-amd64-dev May 14 18:00:54 Kugelfang ferringb|afk: hiya :-) May 14 18:00:54 * Kingtaco gives voice to ferringb|afk May 14 18:01:03 Kingtaco first item: blubb's/dang's multilib issues May 14 18:01:11 Kingtaco blubb/dang, you have the floor May 14 18:01:18 dang blubb: Knock yourself out. May 14 18:01:24 dang I'll fill in any holes May 14 18:01:27 blubb um May 14 18:01:29 blubb ok May 14 18:01:36 Kugelfang hehe May 14 18:01:56 ferringb|afk Kugelfang: yo May 14 18:02:01 blubb well, i assume everybody has read http://dev.gentoo.org/~blubb/meeting May 14 18:02:18 Kugelfang nope, sorry :-) May 14 18:02:36 Kingtaco everyone should take a quick look at it May 14 18:02:47 blubb the libraries that use pkg-config to get the -L parameter aren't a real problem, we just need dang's workaround May 14 18:03:04 blubb the real problem are packages that have their own /usr/bin/*-config May 14 18:03:36 blubb since these scripts are in /usr/bin and not /usr/lib{32,64}, we can't have one per ABI May 14 18:04:15 * JoseJX (~jjezak@josejx.berkshire.net) has joined #gentoo-amd64-dev May 14 18:04:19 * Kugelfang gives voice to JoseJX May 14 18:04:23 * Kingtaco gives voice to JoseJX May 14 18:04:29 blubb so basically we either need to move them somewhere and put a wrapper to /usr/bin, or just modify them to be sensitive to ABI or whatever May 14 18:04:57 ferringb|afk possibility of wrapping them func wise? May 14 18:05:01 Kugelfang blubb: i'd favour your first suggested solution, take them to usr/lib[64] and use a wrapper in /usr/bin May 14 18:05:10 Kingtaco for those that are late joiners, please have a quick look at http://dev.gentoo.org/~blubb/meeting May 14 18:05:12 dang That's what I favor too. May 14 18:05:19 Kugelfang ferringb|afk: -v? May 14 18:05:25 ferringb|afk eg, /usr/bin/pkg-config == whatever the default abi is. for portage innards (ebuild execution), use a func wrapper that is abi sensitive and points at the appropriate binary to use May 14 18:05:29 blubb it's the simple solution May 14 18:05:30 ferringb|afk variation on the symlink approach. May 14 18:05:45 Kugelfang ferringb|afk: that sounds good to May 14 18:05:47 Kugelfang too even May 14 18:05:51 * warpzero (~warpzero@wza.us) has joined #gentoo-amd64-dev May 14 18:05:59 blubb yes, that's another problem May 14 18:06:00 * Kugelfang gives voice to warpzero May 14 18:06:17 blubb when we move from lib32,lib64 to lib,lib64, we would have to change all the modified -configs May 14 18:06:31 dang ferringb|afk: One thing to keep in mind is that we might not want to have a "default ABI" May 14 18:06:39 blubb oh, no, ignore me May 14 18:06:48 Kugelfang blubb: done already:-) May 14 18:06:54 ferringb|afk dang: then you're going to have to modify the *-config to be dependant on $ABI, or hork if it's not there :) May 14 18:06:55 blubb lol May 14 18:06:59 * ferringb|afk is now known as ferringb May 14 18:07:08 Kugelfang ferringb: smaller lines please May 14 18:07:11 dang ferringb: Yes, but it's cleaner. May 14 18:07:15 dang No special cases. May 14 18:07:16 Kugelfang ferringb: the end gets cut off foor me May 14 18:07:20 Kugelfang :-) May 14 18:07:23 ferringb heh, pardon. May 14 18:07:33 ferringb dang: I'd posit it's not. you're suggesting modifying a bunch of ebuilds May 14 18:07:42 ferringb do a wrapper hack (bash function), no modification of *-config is required May 14 18:07:53 ferringb just the "what is the default $ABI" is an issue May 14 18:08:08 blubb that'd be amd64 i guess May 14 18:08:19 blubb btw, is eradicator here? May 14 18:08:19 dang ferringb: The problem is that it makes one of the ABIs second class, which means it'll always have problems. May 14 18:08:29 Kugelfang the bash function can just use access the ABI variable May 14 18:08:30 dang It'll be less tested. May 14 18:08:35 ferringb Kugelfang: exactly. May 14 18:08:44 Kingtaco eradicator said he'd be here May 14 18:08:46 ferringb dang: not really. portage innards will use whatever abi is expected May 14 18:08:48 Kugelfang dang: at last, we're amd64 team, not x86 May 14 18:09:00 ferringb people using *-config outside of portage, they get x86 type behaviour (one to rull them all) May 14 18:09:11 dang Kugelfang: Doesn't mean we can't support 32-bit apps as first class citizens. May 14 18:09:12 ferringb err, sorry Kugelfang :) May 14 18:09:20 dang x86 doesn't matter at all to this, it only has one abi. May 14 18:09:21 blubb ferringb: they should get amd64 May 14 18:09:25 Kugelfang ferringb: length of this was ok May 14 18:09:34 Kugelfang ferringb: but not longer :-) May 14 18:09:37 ferringb blubb: just matters that a choice has to made :) May 14 18:09:49 blubb sure May 14 18:10:14 * Kugelfang votes for feringb's idea May 14 18:10:29 gerrynjr is it time to vote? May 14 18:10:34 ferringb it would require modifying portage initially, but down the line it could be shoved out into an elib. May 14 18:10:37 Kugelfang heh, i guess not May 14 18:10:40 Kugelfang @ gerrynjr May 14 18:10:42 ferringb that's longterm, and assuming I ever get my ass moving. :) May 14 18:10:47 dang I just want to make sure we don't discount a better design because it's harder... May 14 18:11:17 ferringb dang: if the individual *-config become sensitive, they can be exempted. May 14 18:11:18 gerrynjr dang: curious though... on amd64, should the 64 bit abit be the default? May 14 18:11:21 * dang just went through the "quick hack causes problems" at work. May 14 18:11:23 gerrynjr ABI May 14 18:11:32 dang gerrynjr in my opinion, yes. May 14 18:11:40 Kugelfang dang: we'd have to move all those changes upstream May 14 18:11:43 dang On amd64, 64-bit is usually faster. May 14 18:11:50 Kugelfang dang: and they'll say: scew it! May 14 18:11:54 blubb Kugelfang: no, why? May 14 18:12:06 warpzero dang: i haven't found that to be true... May 14 18:12:11 Kugelfang blubb: 2. modify them to be ABI-sensitive. May 14 18:12:11 dang Kugelfang: All we're doing is installing in a different location. May 14 18:12:16 gerrynjr dang: what is the ossue w/ making 32 bit second class then? May 14 18:12:21 dang warpzero: It is in my experience... May 14 18:12:21 Kugelfang dang: RE 2nd solution May 14 18:12:25 Kingtaco 32bit being the default ABI for amd64 makes absolutly no sense May 14 18:12:30 blubb Kugelfang: we could use some ebuild-kungfoo ;) May 14 18:12:33 Kugelfang Kingtaco: righto May 14 18:12:34 gerrynjr Kingtaco: exactly May 14 18:12:44 gerrynjr Kingtaco: I was just getting to that May 14 18:12:54 warpzero i hate to have to ask this, but i just got here May 14 18:12:59 dang Kingtaco: Someone's going to want it. Are we going to say no? May 14 18:13:01 warpzero what is wrong with our current method? May 14 18:13:10 blubb warpzero: it wont work in future May 14 18:13:17 warpzero why not May 14 18:13:19 dang warpzero: We can't do proper dependencies for libraries in both ABIs. May 14 18:13:27 Kingtaco dang, no, we won't deny it, it's just not the default ABI May 14 18:13:48 Kugelfang dang: it's like on ia64 May 14 18:13:49 dang Kingtaco: But it'll be largely untested. May 14 18:14:01 dang Kugelfang: Or sparc, but the other way around. :) May 14 18:14:01 Kugelfang dang: ia64 can execeute x86 and hppa, but the main ABI is ia64 May 14 18:14:27 blubb Kugelfang: that's probably because it can't execute x86 native :P May 14 18:14:37 dang Or fast. May 14 18:14:46 Kugelfang blubb: it can execute it native May 14 18:14:51 blubb oh May 14 18:14:58 gerrynjr in my case though, I run very few 32 bit apps... May 14 18:15:08 Kugelfang blubb: there is a dedicate x86 core in each itanium, but it's clocked at 900-1400MHz May 14 18:15:19 gerrynjr so, i see little use in making 32 bit "default" and It makes more sense to me to have it "second class" May 14 18:15:19 blubb oh, ok May 14 18:15:51 warpzero i think what we're doing now is fine personally, and chroot them if they want more May 14 18:15:58 blubb well, that shouldn't be the big problem May 14 18:16:04 dang warpzero: That doesn't work for end users. May 14 18:16:15 dang You can't tell them "chroot" they won't know how to do it. May 14 18:16:16 warpzero how doesn't it work for end users May 14 18:16:29 warpzero then they probably shouldn't be using gentoo tbqh May 14 18:16:30 dang Portage needs to handle it correctly. May 14 18:16:40 blubb warpzero: what we do now is half multilib half non-multilib May 14 18:16:51 gerrynjr warpzero: that's all to common the attitude.. and it shouldnt be 8( May 14 18:17:07 warpzero blubb: its multilib for specific common libs like glibc right May 14 18:17:11 Kugelfang warpzero: yeah, i _really_ don't care about x86 on amd64 personally, but i'd like to get rid of binary-x86-emul-libs May 14 18:17:14 Kugelfang warpzero: or chroots May 14 18:17:20 blubb oh yeah May 14 18:17:29 Kugelfang warpzero: wouldbe best to let portage handle it May 14 18:17:35 blubb emul-* is evil May 14 18:17:44 cparrott ditto on that May 14 18:17:45 warpzero well then how about the crosscc method May 14 18:17:46 blubb Herbs can confirm that ;) May 14 18:17:57 Kugelfang warpzero: crosscc ? May 14 18:18:04 dang warpzero: It's not building them that's hard, it's tracking them that's hard. May 14 18:18:05 blubb warpzero: crosscc? our gcc builds 32bit code.. May 14 18:18:09 Kugelfang warpzero: the problem is not building them May 14 18:18:10 Kingtaco cross compilation May 14 18:18:18 warpzero thats not what i mean May 14 18:18:25 warpzero i mean having portage make up names on the fly May 14 18:18:36 ferringb eh? May 14 18:18:39 Kugelfang -v ? May 14 18:18:46 * blubb has no idea what warpzero is talking about May 14 18:18:51 * gerrynjr either May 14 18:18:54 warpzero nevermind :/ May 14 18:18:56 * ferringb suspects it's about using abi as slotting for cpv's May 14 18:19:14 ferringb doesn't address the *-config issue originally mentioned however though. May 14 18:19:22 ferringb s:however:: May 14 18:19:37 gerrynjr ferringb: no way to "slot" the *-configs? May 14 18:19:46 gerrynjr ie... have a -config-32 May 14 18:19:50 gerrynjr -config-64 May 14 18:20:02 dang gerrynjr: Not as long as they install in /usr/bin... May 14 18:20:04 Kugelfang gerrynjr: this would be blubb's/dang's first proposal May 14 18:20:15 gerrynjr Kugelfang: oops 8) May 14 18:20:18 ferringb gerrynjr: easiser to jam them in some location, and symlink in the default May 14 18:20:32 ferringb gerrynjr: then (ab|)use bash functions to choose the approriate version based on $ABI May 14 18:20:46 ferringb also provides a route to move away from it, if (fex) pkg-config suddnely supports som $ABI feature May 14 18:20:55 blubb ferringb: that sounds rather hackish to me May 14 18:21:01 warpzero what are other distributions doing May 14 18:21:02 Kugelfang i'd volunteer to add afunction for this to multilib.eclass May 14 18:21:04 gerrynjr *nod* May 14 18:21:08 dang warpzero: Providing binaries. :) May 14 18:21:11 blubb warpzero: nothing afaik May 14 18:21:14 ferringb blubb: eh, it's a careful hack providing a way forward. ;) May 14 18:21:33 gerrynjr well, i'd be interested to see how suse handles somethign like this... May 14 18:21:41 warpzero i seriously vote for chroot and leave it May 14 18:21:45 blubb but if we're moving the -config files anyway, why not make a real wrapper? May 14 18:21:47 gerrynjr I mean.. if I had suse.. how would it deal w/ compiling 32 bit stuffs? May 14 18:21:50 dang gerrynjr: They have different packages for different bitness. May 14 18:21:54 warpzero an architecture is an architecture May 14 18:21:57 gerrynjr blubb: *nod* May 14 18:22:10 blubb gerrynjr: if you have suse, you don't compile. May 14 18:22:13 dang warpzero: And our arch provides two APIs. Shouldn't we support them both? May 14 18:22:16 blubb gerrynjr: not even a kernel, believe me May 14 18:22:20 gerrynjr blubb: i've done it in the past May 14 18:22:31 warpzero dang: it only provides them across a context switch right May 14 18:22:34 cparrott perhaps the right way to think about it is set theory May 14 18:22:38 Kugelfang gerrynjr: that's why you ended up using Gentoo May 14 18:22:38 dang warpzero: Yes. May 14 18:22:41 cparrott x86 is a subset of amd64 as such May 14 18:22:46 cparrott same with ppc and ppc64 May 14 18:22:46 gerrynjr Kugelfang: well, sort of 8) May 14 18:22:55 warpzero its not like ppc and ppc+altivec here May 14 18:23:03 * ferringb is now known as ferringb|afk May 14 18:23:16 Kugelfang Kingtaco: voting on this ? May 14 18:23:50 * Kugelfang already smells the BBW May 14 18:23:52 JoseJX warpzero: ppc + ppc64 isn't ppc and ppc+altivec May 14 18:23:53 Kugelfang BBQ even :-) May 14 18:23:58 blubb heh May 14 18:24:10 warpzero JoseJX: i know thats why i didn't say ppc+ppc64 May 14 18:24:34 blubb Kingtaco: ping? May 14 18:24:35 JoseJX warpzero: Sorry, misunderstood. May 14 18:24:54 dang But amd64 and amd64+SSE3 is a different issue than amd64 +x86, and I believe an already solved on. May 14 18:24:55 dang er one May 14 18:25:01 Kingtaco blubb, pong May 14 18:25:07 blubb ah, he's still alive :) May 14 18:25:09 gerrynjr Kingtaco: voting 8) May 14 18:25:09 warpzero well May 14 18:25:31 Kingtaco ok May 14 18:25:47 Kingtaco let's have 1 line summaries of the proposals May 14 18:25:51 dang warpzero: Of course, chroot won't go away. You can still use it... ;) May 14 18:25:51 Kingtaco then we'll vote May 14 18:26:09 warpzero proposal: do not support x86 on amd64 May 14 18:26:23 blubb we're already doing it May 14 18:26:27 dang == nomultilib May 14 18:26:29 blubb and it's kinda hard-enabled May 14 18:26:36 Kingtaco ok May 14 18:26:39 warpzero so May 14 18:26:41 Kingtaco next proposal? May 14 18:26:43 Kugelfang proposal: use bash function to retrieve correct *-config May 14 18:26:43 warpzero we're talking about what we're going to do in the future May 14 18:27:26 Kingtaco any more proposals? May 14 18:27:28 blubb proposal: use a bash script as wrapper to the real -config May 14 18:27:40 gerrynjr I thought ferringb|afk had one May 14 18:27:48 Kugelfang gerrynjr: i wrote it alreadey May 14 18:27:51 dang gerrynjr: I think that was Kugelfang May 14 18:27:54 dang 's May 14 18:28:08 * Astinus arrives with a bang. May 14 18:28:10 Astinus Sorry I'm late folks May 14 18:28:12 ferringb|afk twasn't I, I Just threw in my 2 cents on it May 14 18:28:14 dang proposal: Rewrite portage from scratch in ruby to do multilib natively May 14 18:28:18 dang :) May 14 18:28:20 blubb lol May 14 18:28:23 Kingtaco ok, so the 2 proposals are, drop multilib support completly May 14 18:28:27 gerrynjr dang: is yours the same as blubbs? May 14 18:28:35 ferringb|afk proposal: hang dang up by his balls and play "beat the pinyada(sp?)" :) May 14 18:28:52 dang gerrynjr: I'm with blubb on this one. May 14 18:29:00 ferringb|afk dang: still wouldn't solve the *-config issue though :) May 14 18:29:03 Kingtaco and write a bunch of bash wrapper scripts to properly support multilib May 14 18:29:28 warpzero okay how about this May 14 18:29:31 blubb well, that's my one 8-) May 14 18:29:40 warpzero proposal: support chrooting for users automatically May 14 18:30:04 blubb warpzero: um? May 14 18:30:04 Kingtaco chroot for build or for runtime? May 14 18:30:14 dang warpzero: Are you going to wrap every binary so it starts in the chroot to run? May 14 18:30:17 Astinus warpzero: That'd mean people essentially need to keep two systems up to date, have two 'full installs' requiring more diskspace, and it's confusing to users May 14 18:30:26 blubb that's a huge overload May 14 18:30:28 dang (remember, with nomultilib, you need to run in the chroot, not just build in it) May 14 18:30:32 warpzero okay, look, May 14 18:30:40 warpzero the way we got it now May 14 18:30:45 warpzero we have a full x86 team May 14 18:30:55 warpzero they debug x86 things May 14 18:31:07 warpzero if we do this wrapper script May 14 18:31:16 warpzero we're going to end up with this secondary library stuff May 14 18:31:25 warpzero where bugs are going to go unnotticed and unfixed May 14 18:31:54 warpzero which is, imho, rediculous May 14 18:31:54 Kugelfang this might be true May 14 18:31:59 gerrynjr bbiab May 14 18:32:01 blubb warpzero: why? the user still is running a amd64-system May 14 18:32:23 Kugelfang blubb: but it means we'd have more bugs to fix May 14 18:32:44 dang Kugelfang: We'll always have more bugs to fix, as long as x86 is the predominant arch on the internet. May 14 18:32:45 * ferringb|afk is now known as ferringb May 14 18:32:48 blubb Kugelfang: well, chances that a tool is broken on x86 but not amd64 are rather low May 14 18:32:54 Kugelfang righto May 14 18:33:03 dang x86-on-amd64 is the least of our bug problems, 64-bit will cause more. May 14 18:33:08 Kugelfang but i mean no x86 bugs May 14 18:33:13 Kugelfang i mean x86 on amd64 bugs May 14 18:33:18 cparrott given that amd64 and x86 (and everything else) use common portage scripts anyway... May 14 18:33:21 Kugelfang dang: right May 14 18:33:32 cparrott ...I would think x86 changes would be noticed for amd64 May 14 18:33:33 * Kugelfang is for voting now May 14 18:33:37 Kugelfang Kingtaco?? May 14 18:33:39 blubb that would be modules-userspace ones, right? May 14 18:33:42 blubb i can't think of any other May 14 18:34:09 Kingtaco ok May 14 18:34:18 Kingtaco so we have 3 proposals right? May 14 18:34:25 Kugelfang yes May 14 18:34:26 Kingtaco 1) drop multilib May 14 18:34:37 Kingtaco 2) write a bunch of wrappers May 14 18:34:55 Kingtaco 3) use chroot for 32bit apps May 14 18:34:59 Kugelfang nono May 14 18:34:59 warpzero 3 isn't valid May 14 18:35:02 warpzero its the same as 1 May 14 18:35:04 dang Kingtaco: You missed one. May 14 18:35:06 Kugelfang 1) drop multilib May 14 18:35:13 Kugelfang 2) write wrapper for -config May 14 18:35:19 dang 3) Portage hack to wrap -config May 14 18:35:21 Kugelfang 2) use bash function to find correct -config May 14 18:35:26 Kugelfang 3) use bash function to find correct -config May 14 18:35:47 blubb Kugelfang: building outside portage would still work, right? May 14 18:35:51 * Astinus is confuddled May 14 18:35:52 dang 1) Punt May 14 18:35:54 warpzero i think that if we're going to fully support multilib May 14 18:35:59 dang 2) hack around in fs May 14 18:36:03 dang 3) hack around in porgage. May 14 18:36:03 Kugelfang blubb: surely May 14 18:36:04 dang :) May 14 18:36:09 warpzero we should force ARCH=x86 to support it to May 14 18:36:15 Astinus Can .. ONE .. person say the proposals we're meant to vote on? :-p May 14 18:36:20 ferringb heh May 14 18:36:22 * Astinus points absently at Taco May 14 18:36:24 Kingtaco agreed May 14 18:36:27 Kingtaco ok May 14 18:36:28 dang warpzero: They do, there's just one lib they can support. :) May 14 18:36:36 Kingtaco 1) drop multilib May 14 18:36:40 warpzero no but i think they should have to use the wrappers May 14 18:36:43 warpzero and stuff May 14 18:36:48 Kingtaco 2) wrappers for -config May 14 18:36:59 Kingtaco 3) fix portage to support multiple abi's May 14 18:37:07 warpzero that way x86 users can switch up "on the fly" May 14 18:37:14 Kugelfang Kingtaco: that 3 is still not correct May 14 18:37:24 blubb lol May 14 18:37:24 Kingtaco ok, what is 3 May 14 18:37:25 warpzero lol May 14 18:37:39 blubb Kingtaco's 3 needs to be done anyway May 14 18:37:41 Kugelfang Kingtaco: 3) write a bash function in portage to automatically select a -config upon ABI May 14 18:37:47 Astinus I take it everyone here is eligible to vote? May 14 18:37:49 Astinus Not just: May 14 18:37:51 Astinus !herd amd64 May 14 18:37:52 jeeves Astinus: (amd64) absinthe, aliz, augustus, eradicator, avenj, config, jhuebel, kugelfang, lv, malc, blubb, slarti, superlag, sekretarz, kingtaco, cryos, luckyduck, voxus, gerrynjr, trapni, r3pek, ticho, flameeyes, astinus, hparker, herbs, tester May 14 18:38:05 Kugelfang Astinus: sorry, no herd, no vote :-P May 14 18:38:07 * Kugelfang hides May 14 18:38:08 dang Astinus: certainly sparc/ppc May 14 18:38:12 blubb hm, can i take back my proposal? it sucks May 14 18:38:18 dang not x86, tho. ;) May 14 18:38:30 Kugelfang blubb: that would be 2? May 14 18:38:32 warpzero thats bs May 14 18:38:34 blubb yep May 14 18:38:36 Astinus Kugelfang: don't you go quickly removing me from herds.xml :P May 14 18:38:37 * JoseJX has quit (Read error: 60 (Operation timed out)) May 14 18:38:39 warpzero what you're about to do affects everybody' May 14 18:38:49 * gerrynjrserver votes 3 May 14 18:38:56 * blubb too May 14 18:39:01 Kingtaco everyone gets a vote May 14 18:39:09 ferringb warpzero: can't get it past the portage crew without it going before the general community. May 14 18:39:14 * Kugelfang votes 3 and volounteers to write it May 14 18:39:14 Kingtaco ok here's how it's going to work May 14 18:39:23 ferringb warpzero: so it's not necessarily a "this is how it shall be", if it involves the portage devs at least. May 14 18:39:31 Astinus Kingtaco: we /msging you our vote? May 14 18:39:31 warpzero :/ May 14 18:39:34 Kingtaco I'm going to call out names, and you respond with 1,2, or 3 May 14 18:39:36 Astinus okay May 14 18:39:45 Kingtaco Astinus May 14 18:39:46 Astinus 3 May 14 18:39:51 Kingtaco avenj, May 14 18:40:08 Kingtaco ... May 14 18:40:11 Kingtaco blubb May 14 18:40:12 blubb 3 May 14 18:40:17 Kingtaco cshields, May 14 18:40:31 Kingtaco ... May 14 18:40:32 * JoseJX (~jjezak@josejx.berkshire.net) has joined #gentoo-amd64-dev May 14 18:40:36 Kingtaco DieMumiee, May 14 18:40:37 * Kugelfang gives voice to JoseJX May 14 18:40:49 Kingtaco ... May 14 18:40:54 Kingtaco eradicator, May 14 18:41:00 Kingtaco .... May 14 18:41:04 Kingtaco Flameeyes, May 14 18:41:05 cshields (sorry, I have to abstain from this one, I don't know the technical details well enough) May 14 18:41:13 Kingtaco cshields, that's fine May 14 18:41:18 Kingtaco voting isn't maditory May 14 18:41:31 Kingtaco gerrynjr, May 14 18:41:35 gerrynjrserver 3 May 14 18:41:40 Tester_ have you all voted ? May 14 18:41:42 Kingtaco Kingtaco: 3 May 14 18:41:49 * ferringb is now known as ferringb|afk May 14 18:41:51 Astinus Tester_: We're voting one by one, as he calls your name May 14 18:41:51 Kingtaco Tester_, we all are now May 14 18:41:57 Kingtaco Kugelfang, May 14 18:41:59 Kugelfang 3 May 14 18:42:04 solar 3 May 14 18:42:04 Kingtaco luckyduck, May 14 18:42:05 luckyduck 3 May 14 18:42:11 * Astinus spanks solar. May 14 18:42:12 Kingtaco mini-morfic, May 14 18:42:21 Kugelfang solar: wait till you're called May 14 18:42:22 Kugelfang :-P May 14 18:42:23 Kingtaco ... May 14 18:42:31 Kingtaco sekretarz, May 14 18:42:42 Kingtaco ... May 14 18:42:46 Kingtaco SpanKY, May 14 18:42:53 Tester_ what's the goal ? May 14 18:43:04 Kingtaco ... May 14 18:43:12 Kingtaco to May 14 18:43:13 SpanKY wtf we voting on ? May 14 18:43:21 Kingtaco 1) drop multilib May 14 18:43:26 Tester_ 1 no May 14 18:43:42 Kugelfang Tester_: read back, 1: drop multilib, 2: use a wrapper script for *-config when building x86 libs, 3: use a portage bash function for finding correct -config script May 14 18:43:43 Kingtaco 2) write bash scripts to patch -config's May 14 18:43:45 solar 1) drop multilib 2) write a bunch of wrappers 3) use chroot for 32bit apps May 14 18:43:52 SpanKY are you kidding me ? May 14 18:43:56 blubb solar: no May 14 18:44:00 dang solar: No, that's not 3 May 14 18:44:02 * gerrynjrserver shakes head May 14 18:44:11 solar yeah thats the wrong one. May 14 18:44:12 dang 1) is use chroot May 14 18:44:22 blubb 3) write a bash function in portage to automatically select a -config upon ABI May 14 18:44:25 Kugelfang <@Kingtaco> 1) drop multilib May 14 18:44:28 Kugelfang 2) wrappers for -config May 14 18:44:32 Kugelfang 2) wrappers for -config May 14 18:44:38 Kugelfang 3) write a bash function in portage to automatically select a -config upon ABI May 14 18:44:45 Tester_ do we really want to build 32bit apps ? May 14 18:44:54 Kugelfang Tester_: not apps May 14 18:44:57 Kugelfang Tester_: libraries May 14 18:44:58 Kingtaco libs May 14 18:45:08 Kugelfang Tester_: this is all about libraries atm May 14 18:45:10 Tester_ arent the binary libraries "good enough" (tm) ? May 14 18:45:14 blubb well, we mainly need those libs for 32bit apps :) May 14 18:45:14 dang 32-bit apps is an entirely different meeting. :) May 14 18:45:26 SpanKY umm, question, how many guys here have worked closely on toolchain multilib ? May 14 18:45:31 SpanKY know anything about what eradicator has been doing ? May 14 18:45:43 Kingtaco eradicator is absent from the meeting May 14 18:45:54 Kugelfang SpanKY: we're talking about how to solve this: http://dev.gentoo.org/~blubb/meeting May 14 18:45:56 Kingtaco which he was specificly requested to be at May 14 18:46:01 Kugelfang SpanKY: so calm down and read it first May 14 18:46:28 Kingtaco SpanKY, I'll come back to your vote May 14 18:46:34 Kingtaco SuperLag, May 14 18:47:05 Kingtaco ... May 14 18:47:08 SpanKY Kugelfang: sorry, but 'drop multilib' was a little too bad sounding :P May 14 18:47:14 Kingtaco Tester_, May 14 18:47:17 Kugelfang SpanKY: spank warpzero May 14 18:47:41 warpzero thats my proposal May 14 18:48:04 Kingtaco everyone is entitled to their idesa as good or bad as they may seem May 14 18:48:04 Tester_ can we just add a prefix to 32bit -config files? May 14 18:48:28 blubb Tester_: no May 14 18:48:34 dang Tester_: Both 2 and 3 do that, they're about how to find the right one. May 14 18:48:46 dang (They put them in /usr/lib[32/64] May 14 18:48:57 blubb erm, yes May 14 18:49:01 Tester_ putting them in /lib is wrong May 14 18:49:01 blubb Tester_: that would be 3 May 14 18:49:01 Kingtaco Tester_, I'll come back to you May 14 18:49:11 Kingtaco voxus, May 14 18:49:12 Tester_ it will break non-portage builds May 14 18:49:21 Tester_ and anyways executables dont belong in /lib May 14 18:49:21 Kingtaco ... May 14 18:49:25 blubb Tester_: we're not talking about that May 14 18:49:31 Kingtaco cparrott, May 14 18:49:35 cparrott 3 May 14 18:49:42 Kingtaco dang, May 14 18:49:44 dang 3 May 14 18:49:49 Kingtaco dmwaters, May 14 18:50:03 Kugelfang Tester_: it's only about how _portage_ should handle it May 14 18:50:11 Kingtaco ... May 14 18:50:15 Kingtaco ferringb|afk, May 14 18:50:29 Kugelfang Kingtaco: ferringb|afk's own proposal was 3 May 14 18:50:48 Kingtaco well, he still has o vote :) May 14 18:50:50 blubb my own was 2 and i voted for 3 :) May 14 18:50:58 Tester_ how would 3 work for calss embedded in configure scripts ? May 14 18:51:01 Kugelfang blubb: bleh May 14 18:51:22 Tester_ s/calss/calls/ May 14 18:51:28 Kingtaco ferringb|afk, I'll come back to you May 14 18:51:31 Tester_ (or in makefiles) May 14 18:51:36 Kugelfang Tester_: setting the appropriate ac_somthing variable before hand May 14 18:51:45 Kingtaco JoseJX, May 14 18:52:24 Kugelfang Tester_: or sed the Makefile or whatever afterwrads using the right -config script's output May 14 18:52:41 Kingtaco ... May 14 18:52:45 Kingtaco lanius, May 14 18:52:45 Tester_ that's messy... May 14 18:53:23 Kingtaco ... May 14 18:53:25 Kingtaco lares_, May 14 18:53:26 Kugelfang Tester_: or: name the function "foo-config", so it get'c called by bash before the real script May 14 18:53:30 Tester_ could we just put the 32bit -configs in a separate dir (lets say /usr/bin/32bitconfigs) and put it first in the path when building 32bit ? May 14 18:53:42 Kugelfang Tester_: shorter line splease May 14 18:53:48 Kugelfang Tester_: i can't read the ne May 14 18:53:50 Kugelfang the end May 14 18:53:57 Kugelfang (friend's computer, don't ask) May 14 18:53:58 Kingtaco .... May 14 18:53:58 Tester_ could we just put the 32bit -configs in a separate dir May 14 18:54:02 Tester_ and put it first in the path when building 32bit May 14 18:54:04 Kingtaco LiveWire, May 14 18:54:28 Kugelfang Tester_: this would still be done by a bash script, wouldn't it? May 14 18:54:46 Kugelfang Tester_: which is what we vote one... using portage for it, or a plain bash wrapper May 14 18:54:48 * gerrynjrserver has quit ("Leaving") May 14 18:54:50 Tester_ true ;) May 14 18:54:54 Kingtaco ... May 14 18:54:58 blubb or dropping multilib ;) May 14 18:54:58 Kingtaco Merlin, May 14 18:55:01 Tester_ well that's portage then ;) May 14 18:55:09 Kingtaco ... May 14 18:55:13 Kingtaco plasmaroo, May 14 18:55:19 Kugelfang Tester_: tell kingtaco :-) May 14 18:55:27 Tester_ Kingtaco: portage all the way ;) May 14 18:55:40 Kingtaco Tester_, that would be #3 May 14 18:55:50 Kingtaco solar, you voted 3 iirc May 14 18:55:55 Tester_ then 3 ;) May 14 18:56:20 Kingtaco suri_, May 14 18:56:27 suri_ 3 May 14 18:56:27 JoseJX Kingtaco: I May 14 18:56:41 Kingtaco JoseJX, k May 14 18:56:46 JoseJX JoseJX: I'm not familiar enough with the issue, I abstain. May 14 18:56:46 * Tester_ feels 3 wins May 14 18:56:48 Kingtaco warpzero, May 14 18:57:06 warpzero whats up May 14 18:57:09 warpzero i can't vote May 14 18:57:15 dang warpzero: Sure you can. May 14 18:57:16 warpzero i'm not on dearest amd64 herd here May 14 18:57:21 Astinus It's open to all ;) May 14 18:57:26 warpzero well okay i vote 1 May 14 18:57:40 Astinus So thats... about 25 votes for '3' and 1 vote for '1' :D May 14 18:57:53 blubb lol May 14 18:58:02 Kingtaco 3 votes for 1 to be correct May 14 18:58:10 warpzero who else voted for 1 May 14 18:58:15 Kingtaco wait, spanky? May 14 18:58:29 Kugelfang Kingtaco: nope May 14 18:58:37 Kugelfang Kingtaco: spanky didn't vote 1 May 14 18:58:49 Kingtaco no, spanky has a chance to vote May 14 18:59:11 Kingtaco er I see, only one vote for 1 May 14 18:59:14 DieMumiee 3 May 14 18:59:15 Flameeyes sorry i had a phone call... anyway 3 is good for me May 14 18:59:20 Kugelfang Kingtaco: is there another item on the agenda ? May 14 18:59:29 Kingtaco yes May 14 18:59:37 Kingtaco but were not quite finished with this one May 14 18:59:44 gerrynjr bbiab, family junk May 14 18:59:46 Kugelfang i'll be back in a few May 14 18:59:55 gerrynjr 5 minutes at max May 14 19:00:04 Kingtaco ok, so we've decided that 3(portage) would be best May 14 19:00:26 gerrynjr actually, we _could_ do it now May 14 19:00:44 Kingtaco now we can either discuss how best to implement it, or go off and work later May 14 19:00:59 gerrynjr (docs) May 14 19:01:07 Kingtaco gerrynjr, it's coming May 14 19:01:07 blubb Kingtaco: i think we should discuss that when eradicator is here too May 14 19:01:14 Kingtaco blubb ok May 14 19:01:20 blubb he's the one that is most familiar with all the multilib stuff May 14 19:01:26 Kingtaco so, let's table multilib for now May 14 19:01:32 Kugelfang kk May 14 19:01:39 blubb and according to his blog he has some plans May 14 19:01:39 Kingtaco we know what we want, we just need to implement it May 14 19:02:04 Kingtaco ok, next item May 14 19:02:12 Kingtaco documentation(amd64) May 14 19:02:17 Kingtaco gerrynjr, you have the floor May 14 19:02:29 gerrynjr our docs atm, have quite a bit of old cruft... May 14 19:02:46 gerrynjr i've also spoken w/ swift and he's stated we need to redoe/revise the install guide May 14 19:03:04 gerrynjr so, if possible i'd like someone to work w/ me on these tasks, May 14 19:03:07 gerrynjr as it's quite a bit May 14 19:03:25 gerrynjr after that, I can get the FAQ up.. May 14 19:03:32 gerrynjr as well as any other doc ideas you may have May 14 19:03:33 * blubb volunteers May 14 19:03:47 blubb although my english sucks ;) May 14 19:03:47 Kingtaco blubb++ May 14 19:03:53 Kugelfang gerrynjr: wanna recruit another amd64/doc developer May 14 19:03:58 dang gerrynjr: I'm not great at writing docs, but I'll help, if someone is willing to polish my work. May 14 19:03:59 gerrynjr Kugelfang: sure 8) May 14 19:04:11 Kugelfang gerrynjr: ok, so let's give this to recruiters May 14 19:04:15 gerrynjr well, the current project site uses guidexml May 14 19:04:16 Kingtaco Kugelfang, let's get lizb back into it May 14 19:04:18 Kugelfang damn, slarti is not here May 14 19:04:23 Kugelfang Kingtaco: righto May 14 19:04:24 Astinus gerrynjr: I can offer some time, but .. it's gonna be 15 days until my finals are over May 14 19:04:29 Astinus Kugelfang: I'll pass it on May 14 19:04:36 Kugelfang Astinus: fine, thx May 14 19:04:44 gerrynjr Astinus: well, i just finshed up, and noticed this pile of docs stuff here, so I'm May 14 19:04:47 gerrynjr wanting to get it done May 14 19:04:49 gerrynjr 8) May 14 19:04:53 * Tester_ can help with the doc, if I find time ;) May 14 19:05:02 gerrynjr Tester_: tat is the limiting factor 8) May 14 19:05:05 gerrynjr that May 14 19:05:07 Astinus gerrynjr: If you want help in 15 days, I'll lend my pitchfork too :) May 14 19:05:24 * cparrott can help too May 14 19:05:25 sekretarz Kugelfang: i have just joined May 14 19:05:39 * [cparrott] (~chris@cpe-66-68-190-228.austin.res.rr.com): Chris Parrott May 14 19:05:39 * [cparrott] +#gentoo-amd64-dev May 14 19:05:39 * [cparrott] irc.freenode.net :http://freenode.net/ May 14 19:05:39 * [cparrott] is an identified user May 14 19:05:39 * [cparrott] idle 00:00:15, signon: Sat May 14 17:19:19 May 14 19:05:39 * [cparrott] End of WHOIS list. May 14 19:05:48 Kugelfang please let us not ask on gentoo-amd64 ML, as we'd get a bunch of people who want to do it, but aren't qulified May 14 19:05:51 Kugelfang sekretarz: :-) May 14 19:06:02 blubb Kugelfang: heh May 14 19:06:10 Kingtaco *snicker* May 14 19:06:18 Astinus Kugelfang: We can put it on the staffing requirements page: Gentoo/AMD64 herd seeks doc monkeys, email gerrynjr May 14 19:06:30 gerrynjr Astinus: sounds good May 14 19:06:31 Kugelfang Astinus: jupp May 14 19:06:58 Astinus Okay, I'll prod someone with commit access there, don't -think- I can fix that for you May 14 19:07:03 gerrynjr initially I went into this thinking there wouldnt be that many docs to revise.. May 14 19:07:08 * Astinus points out dmwaters is here though. May 14 19:07:10 gerrynjr but it seems it significantly more work 8) May 14 19:07:39 Kugelfang ok, /me gone now May 14 19:07:49 * Kugelfang needs food and beer May 14 19:07:58 Kugelfang BBQ!!! May 14 19:08:01 Kingtaco ok, open forum May 14 19:08:01 Kugelfang :-) May 14 19:08:01 Kugelfang later May 14 19:08:04 blubb en guete Kugelfang May 14 19:08:04 Kingtaco l8r May 14 19:08:10 Kugelfang blubb: dank dir May 14 19:08:11 suri_ gerrynjr: Could you give an example of what needs to get done? May 14 19:08:12 Kingtaco FWI May 14 19:08:14 Kugelfang later dudes :-) May 14 19:08:28 Kingtaco FYI: nemo is being upgraded monday May 14 19:08:29 blubb we need to update the technotes about -Os :) May 14 19:09:06 Kingtaco so, I don't expect any data loss, but shit happens... so if anyone has anything they don't want to lose, bleas make a backup before monday May 14 19:09:11 Astinus Hrmmm, how much of this would be better on a 'controlled' style Wiki? May 14 19:09:32 dang Astinus: We got one. May 14 19:09:35 Astinus ie: We give all AMD64 developers access to change shit, and Joe Public can read-only May 14 19:09:38 dang devwiki.g.o May 14 19:09:39 blubb Astinus: not for handbooks and so on May 14 19:09:49 Astinus blubb: Yah, but technotes et al May 14 19:10:02 blubb hm May 14 19:10:02 Astinus dang: Devwiki is dev only May 14 19:10:06 * blubb doesn't like wikis May 14 19:10:20 Astinus blubb: Just thinking from an organizational point of view ;) May 14 19:10:22 warpzero well May 14 19:10:23 warpzero wikis don't like you May 14 19:10:26 * Tester_ thinks the technotes should die some day May 14 19:10:27 blubb then you also could simply give everybody write access to the doc tree May 14 19:10:37 warpzero yeah May 14 19:10:49 blubb Tester_: i agree, but i think we still need some time May 14 19:11:07 Tester_ they were written because I was way to lazy to write a proper install guide ;) May 14 19:11:12 Astinus Just a suggestion 8) May 14 19:11:38 Astinus If we Wiki'd our technotes, we could have a TestedHardware page which is read/write for the public too May 14 19:11:40 blubb btw, as long as i have everybody's attention: PLEASE HELP FIX testing-only.txt May 14 19:11:57 Astinus wtf is testing-only.txt May 14 19:12:01 * Astinus feels out of the loop May 14 19:12:02 warpzero and where is it May 14 19:12:12 Kingtaco http://uberslacks.com/gentoo/amd64/ May 14 19:12:13 blubb http://uberslacks.com/gentoo/amd64/testing-only.txt May 14 19:12:22 gerrynjr suri_: there's info on portions of the project page that iso ut of date, some portions of the nistall guide are no longer relevant, etc.... May 14 19:12:23 blubb a list of packages that have no stable amd64 keyword May 14 19:12:25 Astinus ah, is that our imlate file? May 14 19:12:26 Kingtaco that is testing only and imlate updated every day May 14 19:12:36 blubb i'm running stable amd64 here, and i tell you, it sucks May 14 19:12:47 Astinus HOLY FSCK May 14 19:12:54 warpzero [11:12:48] HOLY FSCK May 14 19:12:55 blubb whenever i want a new app that isn't a depend of gnome, i have to package.keywords whole dependency trees May 14 19:13:02 Astinus testing-only is HUUUUUUUUGE May 14 19:13:09 blubb about 1400 packages, yep May 14 19:13:09 warpzero yes it is May 14 19:13:14 Astinus Erm May 14 19:13:15 Astinus right May 14 19:13:17 Astinus That sucks May 14 19:13:31 blubb and following the policy, we would have to emerge every single package and test it May 14 19:13:35 cshields blubb: is there documentation on how to go about testing/reporting stuff from that file? May 14 19:13:37 blubb that takes some time May 14 19:13:44 Astinus I run ~amd64 and it's wonderfully stable, I switched from just 'amd64' because 'amd64' blows donkey bollocks May 14 19:13:52 Kingtaco cshields, yes May 14 19:13:59 blubb cshields: here it is: emerge it, tell me if it works May 14 19:14:02 Kingtaco take a look at the AT page at amd64.g.o May 14 19:14:08 cshields cool May 14 19:14:12 Astinus Kingtaco: blubb is shortening that policy :P May 14 19:14:16 blubb *g* May 14 19:14:16 Tester_ aren't ATs supposed to do some of the testing for us ? May 14 19:14:25 Kingtaco Tester_, yes May 14 19:14:26 cshields I'll see what I can do.. next weekend my wife and son leave for a week so I'll have some play time May 14 19:14:29 Kingtaco and they have been May 14 19:14:31 Astinus Tester_: Yes, but we don't have enough, they keep becoming Devs :P May 14 19:14:36 * Astinus snickers May 14 19:14:39 Kingtaco the problem is, they keep deving May 14 19:14:52 Kingtaco it's a vicious circle May 14 19:14:56 Astinus Kingtaco: Yah, but they are damn good devs usually May 14 19:15:00 blubb also it would be interesting which packages that you are running ~amd64 and work May 14 19:15:02 gerrynjr Kingtaco: tim to have a AT "lock down" period May 14 19:15:09 gerrynjr time rather May 14 19:15:15 blubb of course only the ones that are on this list May 14 19:15:19 Kingtaco gerrynjr, it's in the works May 14 19:15:24 blubb i'm not really interested in gcc-4 ;) May 14 19:15:24 Tester_ no at's become dev before we cut the backlog by 30% ;) May 14 19:15:29 Astinus blubb: It's a fairly new install, but there's no way I'll change back to stable 8) May 14 19:15:40 Astinus blubb: I'll compare my equery output with that list and send you an email in the week May 14 19:15:51 blubb that'd be great May 14 19:16:05 blubb i'd write a script if i could, but i'm too dumb May 14 19:16:08 Astinus Anyone else running ~amd64 can make it easy that way too 8) May 14 19:16:14 Astinus equery -i (iirc) May 14 19:16:17 gerrynjr i'm running ~amd64 May 14 19:16:26 gerrynjr very stable...