Gentoo Archives: gentoo-dev

From: "Diego Elio Pettenò" <flameeyes@×××××.com>
To: gentoo-dev@l.g.o
Subject: [gentoo-dev] Re: libpng-1.5 smooth upgrade
Date: Fri, 11 Feb 2011 17:27:57
Message-Id: 1297445165.2150.12.camel@raven.home.flameeyes.eu
In Reply to: [gentoo-dev] libpng-1.5 smooth upgrade by "Paweł Hajdan
1 Il giorno ven, 11/02/2011 alle 17.38 +0100, "Paweł Hajdan, Jr." ha
2 scritto:
3 > 1) Are we going to have a tinderbox run *before* libpng-1.5 gets keyworded?
4
5 Absolutely.
6
7 > 2) If the upgrade is non-trivial, i.e. just emerge -uDNa world and
8 > revdep-rebuild isn't going to fix it, will we have an upgrade guide,
9 > possibly as a news item?
10
11 As Samuli said we're planning on making it as "standard" as possible,
12 similarly to what's done with Berkeley DB. This is going to take a bit
13 more work than a standard bump but will avoid further breakage.
14
15 Slotted installation is the thing that makes most sense, especially
16 since upstream helps us there already by versioning the symbols.
17
18 > 3) Can we do something to make catching libpng problems easier? As
19 > Samuli pointed out in https://bugs.gentoo.org/show_bug.cgi?id=354479
20 > there is a predictable compilation warning. How about making portage
21 > print it as a QA warning, so more people can report the issues without
22 > even emerging libpng-1.5 on their systems? That may be a good backup
23 > option for the tinderbox too.
24
25 Tinderbox is also going to report all those warnings to me, which means
26 I'll open them to the bugzilla. While it might not cover 100% use cases,
27 I doubt adding the warning to Portage's "reported" ones is going to
28 help, based on the experience of not even being able to get
29 fortify-sources "will overflow" warnings to be acted upon.
30
31 > Finally, it seems that hard work on --as-needed and automatic fixing of
32 > .la files is going to make the upgrade experience better now, right?
33
34 Most definitely, yes. It could have been better, to be honest, but it's
35 definitely not going to be the many-tiers failure we have seen before.
36
37 --
38 Diego Elio Pettenò — Flameeyes
39 http://blog.flameeyes.eu/