| 1 |
On 14:24 Sat 06 Dec , Alastair Tse wrote: |
| 2 |
[...] |
| 3 |
> However, that python startup time is negligible compared to calling |
| 4 |
> "import portage" where it loads the portdb from disk. For instance, the |
| 5 |
> newest portage calls portageq continuously, and every instance of it |
| 6 |
> does "import portage" which is really slow on my machine. When I say |
| 7 |
> slow, that is a bit more than 2 seconds (5 seconds before disk cache). |
| 8 |
> But if it runs that for every package I have in world, then that is |
| 9 |
> 461*2 ~ 920seconds just for calling "import portage". |
| 10 |
[...] |
| 11 |
|
| 12 |
Unless it creates a new python process every time -- calling "import portage" |
| 13 |
multiple times within a python runtime will only load the portage library once. |
| 14 |
Which, from looking at ebuild.sh it appears to reload the portage library |
| 15 |
constantly (in the case I think you meant to describe), which implies bad design |
| 16 |
and another reason (as drobbins stated) portage's performance hit pertains to a |
| 17 |
reliance on bash. Again, another reason working on portage modularisation |
| 18 |
pertains to greater speed and better design... ;) |
| 19 |
|
| 20 |
-- |
| 21 |
-------------------------~ |
| 22 |
----------Jason-A-Mobarak-~ |
| 23 |
-aether-at-gentoo-dot-org-~ |
| 24 |
-------------------------~ |
| 25 |
|
| 26 |
|
| 27 |
-- |
| 28 |
gentoo-dev@g.o mailing list |
Archives