| 1 |
On Tue, 11 Jul 2017 02:23:56 +0100 |
| 2 |
Ciaran McCreesh <ciaran.mccreesh@××××××××××.com> wrote: |
| 3 |
|
| 4 |
> On Mon, 10 Jul 2017 18:14:23 -0700 |
| 5 |
> Raymond Jennings <shentino@×××××.com> wrote: |
| 6 |
> > If I may ask, does anyone have any profiling information one way or |
| 7 |
> > the other to shed light on the situation? |
| 8 |
|
| 9 |
There are profilers for python. Python devs can comment on such. I am |
| 10 |
not sure about other things like looking for leaks etc. I was not able |
| 11 |
to run python stuff through valgrind. At least I could not run |
| 12 |
java-config through valgrind like I do with jem. |
| 13 |
|
| 14 |
IMHO python all around is just not as robust and should not be used for |
| 15 |
such purposes. There maybe ways to make it faster. For something with |
| 16 |
the complexity, it should really be done in something more robust. Just |
| 17 |
requires talent to make it such. |
| 18 |
|
| 19 |
> Paludis does complete dependency and unused package tracking for |
| 20 |
> everything by default. Any performance difficulties are |
| 21 |
> implementation-related, not a fundamental problem. |
| 22 |
|
| 23 |
I agree its a portage issue. Not saying anyone's code sucks or |
| 24 |
discounting their efforts. Things do not have to be slow. |
| 25 |
|
| 26 |
I mentioned this directly to some portage people in person a few years |
| 27 |
ago during a meeting in Southern California... Around the time I put |
| 28 |
jem on Github. |
| 29 |
|
| 30 |
It is really hard to start over if/when that happens. Thus doing it |
| 31 |
piece meal maybe easier. Though may still end up with spaghetti in the |
| 32 |
end. Hopefully it runs fast and taste good :) |
| 33 |
|
| 34 |
-- |
| 35 |
William L. Thomson Jr. |
Archives