This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author neologix
Recipients gregory.p.smith, neologix, pitrou, sbt, serhiy.storchaka, vstinner
Date 2013-12-08.12:22:58
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <CAH_1eM2v2er4UGoDdtoU8-aHeRZzg_OHxgpkAa_+o92fc2TfpA@mail.gmail.com>
In-reply-to <1386500552.68.0.518343539266.issue19929@psf.upfronthosting.co.za>
Content
> STINNER Victor added the comment:
>
> Since Popen.communicate() returns the whole content of the buffer, would it be safe to increase the buffer size? For example, use 4 GB as the buffer size?

Sure, if you want to pay the CPU and memory overhead of allocating a
4GB buffer :-)

> Should communicate() be "fair" between stdout and stderr?

There's no reason to make the buffer depend on the FD.

> To choose a new value, we need benchmark results on different OSes, at least Windows, Linux, FreeBSD (and maybe also Solaris) since these 3 OSes have a different kernel.

Windows isn't concerned by this, since it doesn't use a selector, but threads.
For the other OSes, that's why I opened this issue (you forgot AIX in
your list :-).
History
Date User Action Args
2013-12-08 12:22:58neologixsetrecipients: + neologix, gregory.p.smith, pitrou, vstinner, sbt, serhiy.storchaka
2013-12-08 12:22:58neologixlinkissue19929 messages
2013-12-08 12:22:58neologixcreate