1 |
On Fri, Dec 2, 2016 at 7:14 AM, Andrew Savchenko <bircoph@g.o> wrote: |
2 |
> Hi all! |
3 |
> |
4 |
> Right now we have two somewhat conflicting policies (at least up to |
5 |
> my understanding of them): |
6 |
> |
7 |
> 1) git atomic commits [1]: |
8 |
> each logical change should be a separate commit. |
9 |
> |
10 |
> 2) revision bump policy [2]: |
11 |
> each change sufficiently affecting application run-time or |
12 |
> installed files should have a revision bump. |
13 |
> |
14 |
> Let's consider the following quite common scenario: package foo-1.0 |
15 |
> should be updated to foo-1.1, but aside from version bump there is |
16 |
> a set of accumulated issues which maintainer is willing to handle, |
17 |
> e.g.: |
18 |
> - bump to EAPI 6; |
19 |
> - fix several runtime bugs (still present in the new version); |
20 |
> - install missing documentation; |
21 |
> - add previously omitted USE flags for some tools of controllable |
22 |
> functionalities; |
23 |
> - etc... |
24 |
> |
25 |
> If both policies are to be followed, users will see something like: |
26 |
> foo-1.0 -> foo-1.1-r8 (assuming each sufficient change was made as |
27 |
> a separate commit with a revision bump). |
28 |
> |
29 |
> While such versioning change is technically correct, it is |
30 |
> confusing for our users and makes future maintainance harder, |
31 |
> because of multiple file renames (yeah, I know about git diff |
32 |
> --find-renames, but this kludge is not perfect). |
33 |
> |
34 |
> What about the following forkflow: |
35 |
> - version bump first with minimal changes required, but without |
36 |
> pushing commit to the tree; |
37 |
> - make each logical change as a separate commit without revision |
38 |
> bumps and without pushing stuff to the tree (of course repoman |
39 |
> scan/full is required as usual for each commit); |
40 |
> - well test package after the last commit (that it builds with |
41 |
> various USE flag combinations, old and new functionality works fine |
42 |
> and so on); |
43 |
> - fix any problems found and only afterwards push changes to the |
44 |
> tree. |
45 |
> |
46 |
> This way users will see only foo-1.0 -> foo-1.1 change in the tree, |
47 |
> while git will still retain each logical change as a separate |
48 |
> commit, which will make future maintenance and debugging much |
49 |
> easier. |
50 |
> |
51 |
> Of course a separate git branch may be used as well, but using |
52 |
> branches for each half-a-dozen set of commits looks like an |
53 |
> overkill to me. |
54 |
> |
55 |
> Thoughts, comments? |
56 |
|
57 |
Thanks for starting the discussion. I completely agree. |
58 |
|
59 |
Though my case might have been a bit more clear-cut since I was |
60 |
working on an ebuild that initially didn't have any KEYWORDS, I think |
61 |
what I did for freeradius is the best way of handling the situation |
62 |
you describe. |
63 |
|
64 |
See 97704b400b7^..e84dc52a816 |
65 |
|
66 |
An initial commit that copied the 3.0.12 ebuild to 3.0.12-r1 without |
67 |
making any other changes, followed by three self-contained |
68 |
fixes/commits, and finally a patch to add KEYWORDS to 3.0.12-r1. |