Gentoo Archives: gentoo-dev

From: hasufell <hasufell@g.o>
To: gentoo-dev@l.g.o, Agostino Sarubbo <ago@g.o>
Subject: Re: [gentoo-dev] Re: [gentoo-commits] repo/gentoo:master commit in: kde-base/kdepim-icons/
Date: Sun, 09 Aug 2015 19:23:34
Message-Id: 55C7A8A7.7080908@gentoo.org
In Reply to: [gentoo-dev] Re: [gentoo-commits] repo/gentoo:master commit in: kde-base/kdepim-icons/ by Agostino Sarubbo
1 On 08/09/2015 08:59 PM, Agostino Sarubbo wrote:
2 > On Sunday 09 August 2015 20:42:00 hasufell wrote:
3 >> So, now we have 21 commits with the exact same commit message:
4 >> =====
5 >> Stable for amd64, wrt bug #556974
6 >> =====
7 >>
8 >> At that point, it's even debatable to have separate commits.
9 >>
10 >> IMO, if you stabilize a huge chunk of ebuilds, you have two possibilities:
11 >>
12 >> 1. just make it one giant commit (we don't revert stabilizations
13 >> anyway)... if it's category-based, start the commit message with
14 >> =====
15 >> kde-base: stable after dev-qt/qtgui bump for ia64
16 >> =====
17 >> if it is related to a particular package (e.g. a bump of dev-lang/ruby
18 >> and very closely related packages that span across multiple categories)
19 >> start it with
20 >> =====
21 >> dev-lang/ruby: stabilize for ia64 including reverse dependencies
22 >> =====
23 >> or somesuch (which is still not very nice, but better)
24 >>
25 >> 2. Have a sensical commit message for each saparate commit. E.g., make a
26 >> local bash hack that automatically prepends "category/pn" to your commit
27 >> message (I think zlogene has done that already) and hope for bug
28 >> https://bugs.gentoo.org/show_bug.cgi?id=557148 to get more attention.
29 >> This is better than mass commits if it can be sensibly automated.
30 >>
31 >>
32 >> Anyway. 21 commits with the same message is really confusing. I can see
33 >> that arch teams might have the most trouble with commit methods, becasue
34 >> they have the highest commit rate, but we have to improve this.
35 >
36 > Since the repoman commit was done per package, I guess the feature requested
37 > in bug 557148 should fix this type of issue.
38 >
39 >
40
41 There's no reason to wait for that. You can automate that
42 with awk/bash/sed whatever too.
43
44 in bashrc e.g.:
45
46 cat_pn() {
47 awk -F/ '{print $(NF-1) "/" $(NF)}' <<<"${PWD}"
48 }
49
50 and then from within the ebuild directories:
51
52 repoman commit -m "$(cat_pn): stable for amd64"