Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #97877 > unrolled thread

select.poll and ppoll

Back to article view | Back to comp.lang.python

Contents

  select.poll and ppoll Steven D'Aprano <steve+comp.lang.python@pearwood.info> - 2015-10-22 16:38 +1100
    Re: select.poll and ppoll Paul Rubin <no.email@nospam.invalid> - 2015-10-22 00:30 -0700
    Re: select.poll and ppoll Laura Creighton <lac@openend.se> - 2015-10-22 11:07 +0200
    Re: select.poll and ppoll Marko Rauhamaa <marko@pacujo.net> - 2015-10-22 14:18 +0300
      Re: select.poll and ppoll Steven D'Aprano <steve+comp.lang.python@pearwood.info> - 2015-10-25 22:30 +1100
        Re: select.poll and ppoll Marko Rauhamaa <marko@pacujo.net> - 2015-10-25 21:21 +0200
          Re: select.poll and ppoll Marko Rauhamaa <marko@pacujo.net> - 2015-10-25 21:23 +0200

#97877 — select.poll and ppoll

Using Python 2.6, don't hate me.

I have select.poll, but I'm looking for something like ppoll instead. From 
the Linux man page:


   ppoll()
       The relationship between poll() and ppoll() is analogous to the
       relationship  between  select(2)  and  pselect(2):  like  pselect(2), 
       ppoll()  allows  an  application to safely wait until either a file
       descriptor becomes ready or until a signal is caught.


Technically, *I* don't want this, it's one of my work-colleagues. He says:

"My high-level goal is to run a callback function whenever the alsa mixer 
level changes.  The C alsa API provides snd_mixer_elem_set_callback, but the 
Python API (import alsaaudio) seems to need me to get poll(2) descriptors"



-- 
Steve

[toc] | [next] | [standalone]

#97879

Steven D'Aprano <steve+comp.lang.python@pearwood.info> writes:
> I have select.poll, but I'm looking for something like ppoll instead. From 
> the Linux man page:

I don't understand your post: do you have a question?

Also I thought the current preferred practice was to use epoll.

[toc] | [prev] | [next] | [standalone]

#97886

In a message of Thu, 22 Oct 2015 16:38:52 +1100, "Steven D'Aprano" writes:
>Using Python 2.6, don't hate me.

>Technically, *I* don't want this, it's one of my work-colleagues. He says:
>
>"My high-level goal is to run a callback function whenever the alsa mixer 
>level changes.  The C alsa API provides snd_mixer_elem_set_callback, but the 
>Python API (import alsaaudio) seems to need me to get poll(2) descriptors"

>-- 
>Steve
>
>-- 
>https://mail.python.org/mailman/listinfo/python-list

Do you need alsa-mixer or would pulse-audio work for you?
Would this: https://github.com/mk-fg/pulseaudio-mixer-cli
do the job?

Laura (alsa-mixer hater)

[toc] | [prev] | [next] | [standalone]

#97893

Steven D'Aprano <steve+comp.lang.python@pearwood.info>:

> I have select.poll, but I'm looking for something like ppoll instead. From 
> the Linux man page:
>
>    ppoll()
>        The relationship between poll() and ppoll() is analogous to the
>        relationship  between  select(2)  and  pselect(2):  like  pselect(2), 
>        ppoll()  allows  an  application to safely wait until either a file
>        descriptor becomes ready or until a signal is caught.
>
> Technically, *I* don't want this, it's one of my work-colleagues. He
> says:
>
> "My high-level goal is to run a callback function whenever the alsa
> mixer level changes. The C alsa API provides
> snd_mixer_elem_set_callback, but the Python API (import alsaaudio)
> seems to need me to get poll(2) descriptors"

I've never used ppoll or pselect. They are used to fix naive signal
handling in a main loop:

    while not signaled:
        select.poll(...)
        ...

The loop suffers from a race condition: the "signaled" flag, which is
set by a signal handler, might change between "while" and "select.poll".

The standard, classic way to solve the race condition is to replace the
"signaled" flag with an internal pipe. The signal handler writes a byte
into the pipe and select.poll() wakes up when the pipe becomes readable.
IOW, ppoll() and pselect() are not needed.


Marko

[toc] | [prev] | [next] | [standalone]

#97941

On Thursday 22 October 2015 22:18, Marko Rauhamaa wrote:

> Steven D'Aprano <steve+comp.lang.python@pearwood.info>:
> 
>> I have select.poll, but I'm looking for something like ppoll instead.
>> From the Linux man page:

[...]

> The standard, classic way to solve the race condition is to replace the
> "signaled" flag with an internal pipe. The signal handler writes a byte
> into the pipe and select.poll() wakes up when the pipe becomes readable.
> IOW, ppoll() and pselect() are not needed.

I don't know who is going to see this, if anyone, since mail.python.org is 
currently down, but thanks for this, and thanks to everyone else who 
answered.

I will discuss with my colleague and see if this solves his problem.



-- 
Steve

[toc] | [prev] | [next] | [standalone]

#97951

Steven D'Aprano <steve+comp.lang.python@pearwood.info>:

> On Thursday 22 October 2015 22:18, Marko Rauhamaa wrote:
>> The standard, classic way to solve the race condition is to replace
>> the "signaled" flag with an internal pipe. The signal handler writes
>> a byte into the pipe and select.poll() wakes up when the pipe becomes
>> readable. IOW, ppoll() and pselect() are not needed.
>
> [...]
>
> I will discuss with my colleague and see if this solves his problem.

Note also that "newer" Linux kernels (2.6.25 and up) have a file
descriptor type, signalfd, that simplifies the idiom slightly.


Marko

[toc] | [prev] | [next] | [standalone]

#97952

Marko Rauhamaa <marko@pacujo.net>:
> Note also that "newer" Linux kernels (2.6.25 and up) have a file
> descriptor type, signalfd, that simplifies the idiom slightly.

<URL: https://pypi.python.org/pypi/python-signalfd>
<URL: https://pypi.python.org/pypi/signalfd>

Haven't tried either.


Marko

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web