#gentoo-qa.irclog

qa.irclog (text/plain), 7.86 KB, created by Kent Fredric (IRC: kent\n) (RETIRED) on 2020-06-14 08:13:27 UTC

[00:00:00] - {Day changed to Saturday, 13 June 2020}
[01:21:01] <kent\n> zlogene: bug #719212
[01:21:04] <kent\n> what information
[01:21:04] <willikins> kent\n: https://bugs.gentoo.org/719212 "sys-auth/pambase calls cpp directly"; Gentoo Linux, Current packages; RESO, WONT; ago:zlogene
[01:21:06] <kent\n> please share
[01:21:54] <zlogene> kent\n: I am not aware it should be open and who told you that, we discussed it before inside qa
[01:22:03] <kent\n> I asked Soap__
[01:22:09] <kent\n> last I checked, he's QA lead
[01:22:36] <kent\n> he literally said it right before I reopened it
[01:22:38] <zlogene> last I checked I had a conversation with him about that
[01:22:43] <zlogene> nothing changed
[01:22:55] <kent\n> was this conversation a decade ago, or in the last hour?
[01:23:01] <kent\n> because mine was in the last hour.
[01:23:12] <zlogene> like 3 weeks ago
[01:23:22] <kent\n> So ... you haven't actually gotten new information.
[01:23:45] <zlogene> kent\n: I could not care more here, sorry
[01:23:54] <zlogene> you do not need to speak for the qa lead
[01:24:07] <kent\n> Just sayin, when I say "QA team lead told me this 30 seconds before I did this", and you say "I have better information", you're calling me a liar.
[01:24:11] <zlogene> as far as I saw him in flesh at least twice
[01:24:17] <zlogene> he had his own tongue
[01:24:20] <zlogene> and head
[01:24:33] <kent\n> How you think any of that is relevant to this conversation is beyond me.
[01:25:01] <kent\n> I even had Soap__ ack the very statement I put forward before I said it.
[01:25:13] <zlogene> kent\n: even qa lead can not step over maintainers' authority
[01:25:41] <kent\n> Um. Ok then.
[01:25:59] <kent\n> But still, you need to retract your "I have better information", because ... you don't.
[01:26:11] <kent\n> You have old information.
[01:26:46] <kent\n> QA lead did in fact say it should stay open. You've chosen not to keep it open in spite of their advice.
[01:26:50] <zlogene> kent\n: do not worry, we can bitch at each other if needed (I mena Soap__ and me) and see what to do ;)
[01:27:12] <kent\n> I mean, its fair enough if there is a good reason not to fix that, but you've not offered any such reason.
[01:27:39] <kent\n> But like, that you wear the QA badge as well seems a bit of an insult to QA right now.
[01:28:03] <kent\n> Because as-is, your implicit statement is "I don't give a fuck about QA"
[01:28:10] <zlogene> kent\n: it is simply preprocessor stuff for if endinf logic, pambase itself compiles nothing
[01:28:15] <kent\n> which, maybe you shouldn't be on the QA team if you don't care about QA.
[01:28:32] <zlogene> kent\n: it is not about qa
[01:28:42] <zlogene> it is all about sanity
[01:29:02] <kent\n> Systems without /usr/bin/cpp are broken, and, you can't assume /usr/bin/cpp is what you think it is.
[01:29:15] <kent\n> that's a bug :p
[01:29:39] <zlogene> kent\n: lots of things are broken without cpp cc string and stuff, the one of them is installed in your system and called gcc
[01:30:16] -*- kent\n is aware GCC breaks due to tc-directly, but its not due to GCC not being able to find other parts of GCC
[01:30:20] <zlogene> but removing semlynks at a whim is like shooting the leg
[01:30:42] <kent\n> its not really "at a whim", its about "simulating different build environments"
[01:30:57] <kent\n> some systems won't have it there.
[01:31:11] <kent\n> we're just synthesizing those systems using a more common one
[01:31:55] <kent\n> and this has potential implications for running prefix systems.
[01:32:02] <kent\n> and cross dev.
[01:32:19] <kent\n> and *thats* why this bug matters.
[01:32:44] <Soap__> it's been settled, this stuff stays
[01:32:54] <Soap__> i.e., USE=native-symlinks
[01:33:44] <kent\n> Soap__: I'm confused now
[01:33:54] <kent\n> you seem to be saying something different to what you said in private.
[01:34:16] <kent\n> I completely agree end users shouldn't have USE="-native-symlinks"
[01:34:25] <kent\n> but all the same, code that breaks with that configuration needs fixing.
[01:34:48] <kent\n> it only shouldn't be fixed, if there's a real problem introduced by doing so.
[01:35:02] <Soap__> kent\n: sorry
[01:35:05] <Soap__> semantic clash
[01:35:09] <Soap__> what I'm saying
[01:35:19] <Soap__> USE="-native-symlinks" as an option needs to stay
[01:35:33] <Soap__> maybe forced on for extra security
[01:36:39] <kent\n> "Forced on" meaning "forced to no-symlinks, or forced to symlink" ? ;p
[01:37:00] <Soap__> forced to symlink
[01:37:11] <Soap__> the fallout currently is too bad
[01:37:14] <kent\n> yeah. And that's fine.
[01:37:27] <kent\n> too much will be broken for users who expect working systems as-is.
[01:37:41] <kent\n> (qt for instance is the biggest problem right now)
[01:38:28] <Soap__> but there needs to be an option to smoke this crap out
[01:38:34] <kent\n> since 2012 when tc-directly was created, the problem has been slowly solved in the bigger part, but its at a point now where most things work as expected, and we can commit to cleaning up the mess with relative effectiveness.
[01:39:27] <kent\n> I mean hell, I'm making perl stuff work under that configuration, I know the pain.
[01:40:06] <Soap__> kent\n: my plan was to make USE="-native-symlinks" obsolete by making all those awful links bash scripts with portage callbacks
[01:40:17] <kent\n> But step 1 is "just make it work enough that a global configuration will work, as long as all other packages build under that same configuration"
[01:40:27] <Soap__> [[ -z ${some_portage_var} ]] && echo "using naughty links"
[01:40:38] <kent\n> Step 2 is the harder part, where you try to make it work with a different potential CC for every package.
[01:41:20] <kent\n> Step 1, is close to being fixed, and I'm not asking for people to try solve Step 2 juusst yet ;p
[01:44:20] <kent\n> zlogene: also, there's no harm in leaving the "depends on" links in place, you can still have a bug "resolved" with that. It just helps people on the cited bug track down things.
[01:44:49] <kent\n> so whatever your opinion is as maintainer on fixing it is, removing that link is still detrimental 
[01:46:16] <zlogene> kent\n: well, as long as David turned on qa bdfl it is not that much you gaing but having it blocked
[01:46:39] <zlogene> s/but/by/
[01:47:47] <kent\n> It just means that people who look at the bug, will be able to easily see the connected bugs. Which they are, absolutely.
[01:48:39] <kent\n> and if QA decides, "yes, this must be fixed", then .... its on them to reopen it, and/or, do it themselves, which last I checked, QA are allowed to do.
[01:58:47] <zlogene> kent\n I have reopened this one
[02:00:26] <kent\n> thanks. And I don't demand it get fixed by you, or anyone, overnight. But like, until it ceases to be a problem, it should be an open bug, if, for no reason other than, so other people who hit the bug in testing can easily search for it
[02:00:55] <kent\n> (otherwise you will get the bug re-reported over and over again in different words )
[02:02:09] <kent\n> perhaps, a feature will land in portage that makes it go away some day, and when that day comes, the bug can be closed without further effort.
[02:02:22] <zlogene> kent\n: I do not think this bug gonna recieve fix at all, becsuse i gonna kill current build system eventually
[02:02:35] <kent\n> that's also fine.
[02:02:53] <zlogene> no idea why Diego picked cpp at all
[02:03:21] <kent\n> If a future incarnation of the ebuild doesn't have this issue, due to a toolchaining replacement, that also "fixes" the bug, even though its a different mechanism
[02:03:57] <kent\n> lol. Diego also made the tc-directly bug. lol
[02:04:45] <zlogene> kent\n as I said, noone notices what pambase calls during its install, everybody gonna notice if i break the configs ;p
[02:26:31] <Soap__> seriously
[02:26:37] <Soap__> CPP has a general macro language
[02:26:45] <Soap__> has to be one of the most asinine things diego did
[02:27:23] <Soap__> as
[02:32:51] <kent\n>  as: command not found'
[02:32:53] -*- kent\n hides