| 1 |
On Tue, 2004-04-06 at 14:13, Chris Smith wrote: |
| 2 |
> On Tuesday 06 April 2004 11:57, Jerry McBride wrote: |
| 3 |
> > Nicholas wrote: |
| 4 |
> > > Ok, This week's project is to test Gentoo's performance... |
| 5 |
> > |
| 6 |
> > How about a "performance wish list" ? |
| 7 |
> > |
| 8 |
> > 1 - Migrate emerge from python to C. |
| 9 |
> > 2 - Use a real database for portage instead of a filesystem |
| 10 |
> > 3 - .... |
| 11 |
> But seriously, folks. Number 2 I can see having good points feature-wise, but |
| 12 |
> not performance. Same with number 1. |
| 13 |
|
| 14 |
Well, it did occur to me, that it would be very useful virtualizing the |
| 15 |
database layer, so that we can use whatever database we want to choose. |
| 16 |
If that's a filesystem based, or a real database, or a REMOTE database. |
| 17 |
|
| 18 |
A remote database would allow managing a horde of gentoo installations |
| 19 |
much easier. It would open up a possiblity of writing a tracker to |
| 20 |
handle security updates, and packages checking on a |
| 21 |
least-change-welcomed corp enviornments. |
| 22 |
This tracker will have to check only one database. |
| 23 |
|
| 24 |
Of course, you can do it right now, only you will have to sync the |
| 25 |
database after every change, or have the tracker poll every host (which |
| 26 |
is abit of the sub-optimal side). |
| 27 |
|
| 28 |
This isn't really "performance", though. |
| 29 |
|
| 30 |
|
| 31 |
|
| 32 |
|
| 33 |
-- |
| 34 |
gentoo-performance@g.o mailing list |
Archives