saghul, on code - Latest Comments

Re: PyStructSequence: creating named tuples in C

Clément Pinard — Thu, 21 Jun 2018 05:35:50 -0000

Hello, thanks for your article !
I hope it's not outdated yet.

I wanted to know how I can convert a struct to a namedtuple in order to pass it in python, and this seems like it could do the job.

However, I did not understand the concept of PyStructSequence_InitType . Why do we need a type for creating a namedtuple ? Shouldn't it be just a PyObject ?

Also what is the line "FooResultType = {0, 0, 0, 0, 0, 0};" doing ? Is it creating a dummy type ? Some documentation tend to say that a PyTypeObject is much longer to specify : https://docs.python.org/3.5...

Thanks in advance for your explanations ! :)

Re: About me

David Cook — Mon, 15 Jan 2018 23:48:17 -0000

well, i guess i will be upfront...i'm looking for a developer to ugrade my service. if you are interested or know someone who is, please contact me at the above address, thank you.

Re: About me

David Cook — Wed, 03 Jan 2018 20:22:29 -0000

hello, can you contact me at dave@copycall.com
thank you.

Re: lambda vs. functools.partial

Adam Brzęcki — Wed, 13 Dec 2017 11:58:14 -0000

Amazing article!
Using "partial" fixed my problem. Before I was trying to implement "lambda" but the problem was with "late binding". My lambda expression (anonymous function) was executed each time I was pressing button (using PyQt 4) and that expression remembered only last value. I was searching another way to "tattoo" name of function to call and value of parameter into my button object - not to load it each time I press the button. "Partial" allowed me to tattoo this costruction into my button. Thanks! :)

Re: Implementing registry pattern with class decorators

saghul — Fri, 15 Sep 2017 16:50:53 -0000

It depends on the python-application package, which I forgot to mention!

Re: Implementing registry pattern with class decorators

Brett Swift — Fri, 15 Sep 2017 14:33:49 -0000

where does application.python.util.Singleton come from? It doesn't work to have sample code if it's not all present.

Re: Winds of change

Oier — Wed, 21 Dec 2016 07:45:10 -0000

Congrats man! You deserve it, so have fun and hardwork in your new beggining ;).

Re: Winds of change

mikjimenez — Wed, 21 Dec 2016 05:49:36 -0000

We all know this new chapter in your career is gonna be awesome and you gonna kick some ass ;)

Congrats! change always has positive things ;)

Re: LXD, Debian containers and systemd

brontide — Sun, 23 Oct 2016 12:10:46 -0000

I spent several hours running up against this before filing a bug and having them close it within minutes is so infuriating. If debian jessie is broken under lxc they should not be publishing images based on it. I'll try this recipe to see if this solves my problems as well.

Re: LXD, Debian containers and systemd

Kim Callis — Wed, 20 Jul 2016 18:45:18 -0000

I have been trying to be somewhat creative in doing Debian 8 under Ubuntu 16.04. The creative trick is trying to install Xivo (PBX in a Flash). I created the container, but I can't seem to have an environment that I would have had is I were doing netinstall....

Re: Trying out concurrent.futures

San Luthra — Sat, 09 Aug 2014 11:04:58 -0000

The reason for the performance difference is that multiprocessing.Pool will batch the iterable passed to map into chunks, and then pass the chunks to the worker processes, which reduces the overhead of IPC between the parent and children. ProcessPoolExecutor always passes one item from the iterable at a time to the children, which can lead to much slower performance with large iterables, due to the increased IPC overhead. However, looking at this message: http://bugs.python.org/issu... It looks like using concurrent.futures is not bad choice if utterable is small and if you use the latest API hopefully in Py3.x if the API are improved if the put multiprocess into concurrent/* and you are on the right track.

Re: Using functools.partial instead of saving arguments

saghul — Tue, 26 Nov 2013 10:39:06 -0000

Interesting :-) Still, saving a stack frame is nice for traceback printing ;-)

Re: Using functools.partial instead of saving arguments

schlamar — Tue, 26 Nov 2013 10:19:45 -0000

Pypy 2.1 (please note that the order of the tests is changed)

In [2]: %timeit testpartial.do_test(testpartial.WorkItemPartial)

1000000 loops, best of 3: 535 ns per loop

In [3]: %timeit testpartial.do_test(testpartial.WorkItem)

10000000 loops, best of 3: 58.5 ns per loop

Re: Serving a WSGI app, WebSockets and static files with Twisted

zanio — Fri, 11 Oct 2013 11:01:24 -0000

great, thanks

Re: On the libuv core team!

mikjimenez — Fri, 06 Sep 2013 21:49:22 -0000

Congratsss!

Re: Using functools.partial instead of saving arguments

Juanlu001 — Fri, 06 Sep 2013 18:54:39 -0000

And apparently it saves you a stack frame!

http://tech.pro/tutorial/15...

Re: Trying out concurrent.futures

saghul — Fri, 06 Sep 2013 18:46:57 -0000

To be honest I never used multiprocessing.Pool myself :-S Not sure how much overhead Futures add, here, but you might have found a nasty bug, they should perform equally, I guess.

Re: Trying out concurrent.futures

Juanlu001 — Fri, 06 Sep 2013 18:45:47 -0000

I made a change so now they look almost identical, and the performance of pool_futures.py is still very poor:

https://gist.github.com/Jua...

Re: Trying out concurrent.futures

Juanlu001 — Fri, 06 Sep 2013 18:38:14 -0000

I don't fully understand the advantages of using `ProcessPoolExecutor` versus the old `Pool` in multiprocessing. Consider these two pieces of code, adapted from yours:

https://gist.github.com/Jua...

In my computer:

[juanlu@nebulae]─[~/Development/Python/test]
└[00:35:30] $ time python pool_multiprocessing.py

real 0m0.190s
user 0m0.217s
sys 0m0.020s
[juanlu@nebulae]─[~/Development/Python/test]
└[00:35:31] $ time python pool_futures.py

real 0m2.426s
user 0m3.400s
sys 0m0.577s

I didn't understand the advantages, and now I understand the performance even less. Trying to run cProfile on them gives strange errors, so I couldn't figure out the issue. What do you think?

Re: On the libuv core team!

Juan Riaza — Fri, 06 Sep 2013 17:38:10 -0000

Congratz!!!

Re: uvent: a gevent core implemented using libuv

blog.est.im — Sun, 03 Mar 2013 20:18:26 -0000

damn you are so productive.

Re: Fast(er) locks in Python?

saghul — Sun, 11 Nov 2012 12:09:43 -0000

Thanks for pointing it out Antoine! I added an update to the post.

Re: Fast(er) locks in Python?

Antoine — Sun, 11 Nov 2012 08:17:58 -0000

For the record, threading.RLock is now implemented in C (since Python 3.2, I believe), and the timeout option also is.

Re: Using functools.partial instead of saving arguments

João Mesquita — Sat, 10 Nov 2012 07:49:25 -0000

It is a bit less readable for the not so fond of functional programming right? I guess it is a trader off between code complexity and performance.

Re: uvent: a gevent core implemented using libuv

Gonçalo — Thu, 04 Oct 2012 22:33:05 -0000

Two questions, please:1) Is it possible to implement libev on top of libuv with uvent as abstraction layer between them?2) Does it work in Firefox? I need it for a project (an already sold project).THANKS!