Hello everybody, I am Severin - a student at ETH in Zurich. Currently i'm enrolled at the Masters program in computer science. The next 6 month, starting from the 15th of September I will do my thesis. This more or less means working on a project. Since me and my mentor (Nicholas Matsakis) both share interest in python we came up with the idea to do some work with the pypy project. We started to talk about what we could do and what would be interesting for us. In some way we wanted to '*improve multiprocessor capabilities*' in pypy. We had some ideas so far, but we are not yet fixed on something. I was wondering if somebody of you has something that is of particular interest to the pypy community. Some ideas that might be interesting to look at for me. Any kind of input or idea is welcome... Best regards, Severin
2008/8/27 Severin <seob@gmx.ch>:
We started to talk about what we could do and what would be interesting for us. In some way we wanted to 'improve multiprocessor capabilities' in pypy. We had some ideas so far, but we are not yet fixed on something.
I was wondering if somebody of you has something that is of particular interest to the pypy community. Some ideas that might be interesting to look at for me. Any kind of input or idea is welcome...
A random idea. Implementing python-safethread monitors for PyPy? http://code.google.com/p/python-safethread/wiki/Monitors -- Seo Sanghyeon
safethread sound like something interesting. are there already any safthread ideas in pypy? In general, working on GILless ideas sounds more interesting than doing something on multiprocess framworks. It's a master thesis, that means there is room for experimental stuff, but in the end half a year is quite short, so the room is limited again :) was there any attempt to remove the GIL in pypy so far? -- Severin On Tue, Aug 26, 2008 at 6:21 PM, Seo Sanghyeon <sanxiyn@gmail.com> wrote:
We started to talk about what we could do and what would be interesting for us. In some way we wanted to 'improve multiprocessor capabilities' in
2008/8/27 Severin <seob@gmx.ch>: pypy.
We had some ideas so far, but we are not yet fixed on something.
I was wondering if somebody of you has something that is of particular interest to the pypy community. Some ideas that might be interesting to look at for me. Any kind of input or idea is welcome...
A random idea. Implementing python-safethread monitors for PyPy? http://code.google.com/p/python-safethread/wiki/Monitors
-- Seo Sanghyeon
On Wed, Sep 3, 2008 at 11:31 AM, Severin <seob@gmx.ch> wrote:
safethread sound like something interesting. are there already any safthread ideas in pypy? In general, working on GILless ideas sounds more interesting than doing something on multiprocess framworks. It's a master thesis, that means there is room for experimental stuff, but in the end half a year is quite short, so the room is limited again :) was there any attempt to remove the GIL in pypy so far? -- Severin
None that I know of at least. I think any serious GIL-removal proposal requires some deeper thinking about what exactly are the expected semantics. Cheers, fijal
On Wed, Sep 3, 2008 at 3:44 AM, Maciej Fijalkowski <fijall@gmail.com> wrote:
On Wed, Sep 3, 2008 at 11:31 AM, Severin <seob@gmx.ch> wrote:
safethread sound like something interesting. are there already any safthread ideas in pypy? In general, working on GILless ideas sounds more interesting than doing something on multiprocess framworks. It's a master thesis, that means there is room for experimental stuff, but in the end half a year is quite short, so the room is limited again :) was there any attempt to remove the GIL in pypy so far? -- Severin
None that I know of at least. I think any serious GIL-removal proposal requires some deeper thinking about what exactly are the expected semantics.
Oi, how did I miss this thread before... Python-safethread's design of course includes most of this deep thinking. -- Adam Olsen, aka Rhamphoryncus
Hi, On Wed, Sep 03, 2008 at 12:15:37PM -0600, Adam Olsen wrote:
Python-safethread's design of course includes most of this deep thinking.
Jython includes another possible outcome of deep thinking about this. In general there are various possible approaches. A bientot, Armin.
Hello. Can you explain a bit in which direction would you like to go? Or is the topic vague for a reason to experiment a bit? Would you rather want say, create some GILless locking mechanism in pypy or rather implement some cool multi-process framework based on what is done already? Cheers, fijal On Tue, Aug 26, 2008 at 5:35 PM, Severin <seob@gmx.ch> wrote:
Hello everybody,
I am Severin - a student at ETH in Zurich. Currently i'm enrolled at the Masters program in computer science. The next 6 month, starting from the 15th of September I will do my thesis. This more or less means working on a project. Since me and my mentor (Nicholas Matsakis) both share interest in python we came up with the idea to do some work with the pypy project.
We started to talk about what we could do and what would be interesting for us. In some way we wanted to 'improve multiprocessor capabilities' in pypy. We had some ideas so far, but we are not yet fixed on something.
I was wondering if somebody of you has something that is of particular interest to the pypy community. Some ideas that might be interesting to look at for me. Any kind of input or idea is welcome...
Best regards, Severin
_______________________________________________ pypy-dev@codespeak.net http://codespeak.net/mailman/listinfo/pypy-dev
participants (5)
-
Adam Olsen -
Armin Rigo -
Maciej Fijalkowski -
Seo Sanghyeon -
Severin