| 1 |
On 00:10 Sat 06 Dec , Aron Griffis wrote: |
| 2 |
> Bob Miller wrote: [Fri Dec 05 2003, 01:46:10PM EST] |
| 3 |
> > Second, is there anything wrong with Python as an implementation |
| 4 |
> > language? If you think Python is too slow, think again. On nearly |
| 5 |
> > every portage operation I do, the CPU is mostly idle -- it's the disk |
| 6 |
> > that's thrashing (according to gkrellm). The key to improving |
| 7 |
> > portage's performance is to get it to open fewer files. |
| 8 |
> |
| 9 |
> Sort of. It's true that there's a lot that could be done to improve |
| 10 |
> portage performance without changing languages. Nonetheless, even the |
| 11 |
> smallest test shows that python has a poor startup time. |
| 12 |
> |
| 13 |
> $ time python -c 'print "hi"' |
| 14 |
> user 0m0.023s |
| 15 |
> sys 0m0.004s |
| 16 |
> |
| 17 |
> $ time perl -e 'print "hi\n"' |
| 18 |
> user 0m0.003s |
| 19 |
> sys 0m0.003s |
| 20 |
[...] |
| 21 |
|
| 22 |
I'd like to point out that that start up time is negligible -- especially since |
| 23 |
type typical emerges aren't going to be hindered by a less than one half second |
| 24 |
start up time. Part of the reason I am working on modularisation of portage is |
| 25 |
that you don't have to pull 5000 lines of python code along with a 50 line |
| 26 |
utility script. |
| 27 |
|
| 28 |
Also, if a bash script is feeling a hit from constantly reloading the portage |
| 29 |
library, it probably needs to be rewritten in python. |
| 30 |
|
| 31 |
-- |
| 32 |
-------------------------~ |
| 33 |
----------Jason-A-Mobarak-~ |
| 34 |
-aether-at-gentoo-dot-org-~ |
| 35 |
-------------------------~ |
| 36 |
|
| 37 |
|
| 38 |
-- |
| 39 |
gentoo-dev@g.o mailing list |
Archives