1 |
On Thu, Mar 01, 2012 at 11:45:40AM +0300, Alexey Shvetsov wrote: |
2 |
> Not sure if its for council. |
3 |
I don't think it's suitable for council, but it does need more |
4 |
discussion than it's been getting on the scm list. |
5 |
|
6 |
> Define timeframe for cvs->git migration. Technical issues almount |
7 |
> finished |
8 |
> 1. Thin manifests already in portage |
9 |
> 2. Git commit signing already there (if git-1.7.9 will be keyworded due |
10 |
> to issue with svn 1.7 [git-svn functionality with SVN 1.7 if your |
11 |
> SVN repo URL, branch name or tag names contains characters that need |
12 |
> URL escaping]) |
13 |
Do you have a fix for the Git SVN 1.7 problem then? |
14 |
|
15 |
Other items not included in the above, that we need some agreement on. |
16 |
- Initial tree state: |
17 |
Single commit, signed by me, with a graft of history available as a |
18 |
separate download (it's ~1200MB) . This ensures that all of the |
19 |
history is available AND that the usual downloads for developers are |
20 |
very small. |
21 |
Alternatively, how much history should we include in the base |
22 |
download? |
23 |
- Log generation |
24 |
- Potentially dropping Changelogs in Git (generate |
25 |
shortly during rsync tree generation?) |
26 |
- Merge policies |
27 |
This is the hardest political topic. |
28 |
Do we force users to rebase before they push to the tree? |
29 |
Before you say yes, of course, there is a catch: |
30 |
If the user publishes their work in MORE than one place, and does push |
31 |
to remote A/master, pull from remote B/master, rebase master on |
32 |
B/master, push to remote B/master. Now what do they do about the state |
33 |
of remote A? They _HAVE_ to either have a merge of A/master, or |
34 |
destroy the history at A/master with a forced push. |
35 |
|
36 |
-- |
37 |
Robin Hugh Johnson |
38 |
Gentoo Linux: Developer, Trustee & Infrastructure Lead |
39 |
E-Mail : robbat2@g.o |
40 |
GnuPG FP : 11ACBA4F 4778E3F6 E4EDF38E B27B944E 34884E85 |