1 |
-----BEGIN PGP SIGNED MESSAGE----- |
2 |
Hash: SHA1 |
3 |
|
4 |
Donnie Berkholz wrote: |
5 |
> I'm still not convinced of that. The alternative would require a patch |
6 |
> to git-submodules, some additional tools support for grabbing additional |
7 |
> repos with checkouts and repoman, but it has some compelling advantages |
8 |
> (like non-devs don't need 1 GB of web space / bandwidth to post their |
9 |
> forks of the tree). |
10 |
|
11 |
Aren't the existing overlays a suitable method of tree forking? |
12 |
|
13 |
I agree, 1Gb's a bit steep, but someone commented[1] that they'd |
14 |
compressed their (relatively current) checkout down to 650ish Mb. I'm |
15 |
currently attempting a complete import using git-cvsimport (from an |
16 |
rsynced copy, don't worry Robin!). It's still updating (so far 15+ |
17 |
hours, so I hope I haven't made a mistake back at the start!), and it'll |
18 |
need repacking after that, but then it'll hopefully be relatively easy |
19 |
to add incremental updates. |
20 |
|
21 |
Rereading past threads it looks like we still need someone to try |
22 |
setting up and running repoman against git's cvsserver implementation. |
23 |
|
24 |
I'll see how small I can make my repo, and if it's significantly smaller |
25 |
(under 700Mb), I can post it to dev.g.o for those that want to duplicate |
26 |
it (although, as ever, please be nice to infra and yoink it using http |
27 |
not ssh). Also remember that doing cvsimported git repos should be |
28 |
cloned from rather than modified directly (I think it breaks the ability |
29 |
to do further incremental updates)... |
30 |
|
31 |
Mike 5:) |
32 |
|
33 |
[1] http://kerneltrap.org/mailarchive/git/2008/12/13/4410084/thread |
34 |
|
35 |
-----BEGIN PGP SIGNATURE----- |
36 |
Version: GnuPG v2.0.9 (GNU/Linux) |
37 |
|
38 |
iEYEARECAAYFAkl9/J0ACgkQu7rWomwgFXpo5QCfSYF5hDJj++rVCdAz2lO6CqxT |
39 |
EJsAn0iwd7brHGiZduol5ciGF6TtnvT3 |
40 |
=CS+a |
41 |
-----END PGP SIGNATURE----- |