| 1 |
On Wed, Feb 18, 2009 at 01:24:20PM -0800, Donnie Berkholz wrote: |
| 2 |
> What are some other options? We could have a manifest directory instead, |
| 3 |
> with 1 file inside per distfile. |
| 4 |
Please no. The space savings that came from using a single Manifest |
| 5 |
rather than the split digests per ebuild were significant. |
| 6 |
|
| 7 |
> > Hence the question - is it possible to *not* store and .gitignore Manifests is |
| 8 |
> > git controlled portage repository? |
| 9 |
> > As portage metadata is regenerated, maybe it would be as well possible to |
| 10 |
> > regenerate manifests on server? |
| 11 |
> > I guess it would be possible but ineffective as it would require all needed |
| 12 |
> > distfiles to be present as well and this is unacceptable. |
| 13 |
The problem just needs solving from a different angle. |
| 14 |
|
| 15 |
If you look at the existing Manifests, there are 4 entry types: |
| 16 |
AUX |
| 17 |
EBUILD |
| 18 |
DIST |
| 19 |
MISC |
| 20 |
|
| 21 |
DIST is the only one that needs something outside the directory to |
| 22 |
generate. It's also the most time-consuming to generate. |
| 23 |
|
| 24 |
Using the converse, all files covered by AUX, DIST, MISC have GIT SHA1 |
| 25 |
commit ids. Explicitly performing a checksum on them is not needed, just |
| 26 |
extract it from Git. |
| 27 |
|
| 28 |
A distfile-only Manifest should merge extremely well, as it will seldom |
| 29 |
have conflicts. |
| 30 |
|
| 31 |
When it comes to generating the outgoing Manifests for users on the |
| 32 |
central server, it's pretty simple. |
| 33 |
|
| 34 |
The only downside I see is the potential for a degree of lesser security |
| 35 |
for anybody using the Git repo directly instead of rsync. |
| 36 |
|
| 37 |
-- |
| 38 |
Robin Hugh Johnson |
| 39 |
Gentoo Linux Developer & Infra Guy |
| 40 |
E-Mail : robbat2@g.o |
| 41 |
GnuPG FP : 11AC BA4F 4778 E3F6 E4ED F38E B27B 944E 3488 4E85 |
Archives