[17:01] <mdeslaur> \o
[17:01]  * stgraber waves
[17:01]  * mdeslaur waves to Schrödinger's tech board
[17:02] <pitti> o/
[17:03] <pitti> there's some interesting discussion on u-devel@ about archive reorg episode VII
[17:03] <pitti> otherwise I don't see much of an agenda
[17:04] <mdeslaur> I like the proposal
[17:05] <pitti> TBH it sounds to me like trading in some short-term convenience for piling even more on top of our ever-growing tech debt long-term
[17:05] <pitti> "I can't build ubuntu touch because of that half-done Haskell transition"
[17:05] <pitti> and the like
[17:06] <pitti> but oh well, if people want that pain, so be it
[17:06] <kees> o/
[17:07] <mdeslaur> pitti: hrm, I don't quite understand that example
[17:07] <pitti> but the thing that'll be an absolute disaster is to enable universe for image builds and (try to) disable it again at release
[17:07] <pitti> this will sooo not work
[17:07] <mdeslaur> pitti: oh, that part will never work, yeah
[17:07] <pitti> mdeslaur: well, half of the time the Haskell stack or other bits are uninstallable because of half-done transitions
[17:08] <pitti> mdeslaur: if we now start build-depending on that for a lot of crucial packages, we'll suddenly find ourselves in a position where we have to fix the entire <censored> Haskell transition (which happen like every month) before -proposed becomes installable again
[17:08] <pitti> mdeslaur: just look at http://people.canonical.com/~ubuntu-archive/component-mismatches.svg
[17:08] <pitti> and imagine -- one uninstallable package in that mess will ruin your day
[17:09] <pitti> right now it ruins some universe packages, but *shrug*, they'll just stay in -proposed until someone cares, or get kicked out
[17:09] <mdeslaur> ah, I see what you mean...we now have to care about all the universe packages
[17:09] <pitti> because all of a sudden there is no boundary to what we call "supported" officially
[17:10] <mdeslaur> which kind of defeats the purpose of trying to care less
[17:10] <pitti> that's the kind of long-term trade-off that I mean
[17:10] <pitti> it might *seem* easier, but in half a year we might find it's a complete loss
[17:10] <pitti> I believe that the MIR process *has* to be painful and long
[17:10] <pitti> we already have way too much crap in main  which people never get rid of
[17:10] <pitti> thus blowing up our images, security support, etc.
[17:11] <mdeslaur> yeah, I now see the issue
[17:11] <pitti> i. e. if you want to pull in a new toolchain to build your package, I think the onus should be on *you* to reconsider if you don't rather use the ubuntu "standard" technology/API
[17:11] <pitti> instead of "someone in the release team will figure it out two days before release"
[17:11] <pitti> so that kind of barrier is not a bug IMHO, it's a feature
[17:12] <pitti> and if anything they aren't high enough :)
, sorry
[17:12] <pitti> didn't mean to turn this meeting into a soapbox
[17:12] <mdeslaur> no, I think that makes sense, and it's something I had not thought about
[17:12] <mdeslaur> trying to get rid of a 5-year support commitment by opening up universe then makes universe something we have to support at least until release
[17:13] <mdeslaur> hm
[17:13] <pitti> well, if we don't want to support universe, we'd have to use less of it, not more..
[17:15] <pitti> anway, no quorum, no official board, I suppose we skip the official meeting?
[17:15] <mdeslaur> so what we really need to something between main and universe, a buildmain or something
[17:15] <mdeslaur> yeah, I think we can skip
[17:15] <pitti> right, I proposed a main-build-deps component in the middle
[17:15] <mdeslaur> kees: did you have anything?
[17:15] <pitti> but that was rejected
[17:18] <pitti> stgraber: anything from your side?
[17:18] <pitti> (sorry, we actually do have quorum!)
[17:20] <stgraber> I did read the proposal back when it was still a WIP, haven't had time to follow much of the discussion since though
[17:21] <pitti> I'll follow up on the ML again wrt. the main-build-deps component
[17:21] <kees> mdeslaur: nothing from me
[17:22] <stgraber> tbh, for a while now the biggest pain with main promotions has been the multi-month delay for security reviews
[17:22] <pitti> right
[17:22] <stgraber> to the point where we've had to promote things basically on release week because it took so long to review, that resulted in a bunch of breakage in wily which we've had to deal with through SRUs
[17:22] <pitti> but that's not going to go away for things where a security review is actually relevant and requested
[17:22] <stgraber> that's also not something I'd expect to change with that bit of archive reorg
[17:23] <pitti> multi-month is certainly the issue which needs to be addressed
[17:23] <pitti> two weeks or so would be fine, but of course if there's too much demand for new stuff that should increase the pressure to maybe take a step back and ask "do I really need that" :)
[17:24] <stgraber> we could workaround the issue by having the security review be non-blocking. That is the MIR team requests the review by filing a separate critical bug on the package, but let the promotion go ahead regardless.
[17:28] <pitti> that sounds workable
[17:28] <pitti> if the MIR gets a new point "contingency plan if security review outcome is negative"
[17:28] <pitti> i. e. how the feature can be pulled without disrupting half of ubuntu
[17:29] <pitti> I just followed up to the ML wrt. the intermediate component
[17:29] <pitti> stgraber: want to follow up about the "speed up MIR reviews" proposal?
[17:30] <stgraber> overly busy right now but I may look into that next time I'm stuck on security review and get frustrated :)
[17:30] <mdeslaur> hehe
[17:36] <pitti> ok, I think we're done here?
[17:37]  * pitti waves good night
[17:37] <stgraber> sounds like it
[17:38] <mdeslaur> thanks everyone
[17:38] <pitti> cheers