1 |
On Tue, 22 Apr 2003 21:56:16 -0500 |
2 |
Brian Jackson <brian@××××.com> wrote: |
3 |
|
4 |
> I know talking to myself this much in one day is bad, but I setup an rsync |
5 |
> server. It is at: |
6 |
> rsync://www.mdrx.com/portage-local |
7 |
> |
8 |
> you can run the following to check it out: |
9 |
> rsync -rz rsync://www.mdrx.com/portage-local . |
10 |
|
11 |
Actually, more formal support for multiple sync sources would be *highly* |
12 |
cool & would really make portage a "meta-distribution". I could also see this |
13 |
being useful for officially supported builds, as this readily solves the "I |
14 |
never want to even sync anything X11" problem. |
15 |
|
16 |
On a (barely) related extensibility note, I've found it really hard to work on |
17 |
even short patches to portage/emerge. portage.py is almost 4600 lines, emerge |
18 |
is almost 1900. My point being that's a pretty good size for a single code unit |
19 |
in any language, but enormous for python. Try |
20 |
|
21 |
find /usr/lib/python2.2/ -iname "*.py" |xargs wc -l |sort -n -r |
22 |
|
23 |
and keep in mind that the standard modules have a lot more documentation. And |
24 |
to cut off the inevitable criticism of this metric, these modules do import each |
25 |
other & use C libs. That's my point. The lack of modularity & the presence of |
26 |
top-level code makes reuse really tough- I'd love to work on a GTK'd or web'd |
27 |
emerge or 30 other scripts, but everything's so intertwined that it's not |
28 |
feasible to reuse code. |
29 |
|
30 |
More user whining, I know. I don't quite know what I'm suggesting as a |
31 |
solution, short of offering to help. |
32 |
|
33 |
--Pete |
34 |
|
35 |
-- |
36 |
Peter Fein |
37 |
pfein@×××××.com |
38 |
773-575-0694 |
39 |
|
40 |
-- |
41 |
gentoo-dev@g.o mailing list |