Author gregory.p.smith
Recipients gregory.p.smith, gvanrossum, neologix
Date 2013-12-01.19:40:23
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1385926824.0.0.338852467339.issue19850@psf.upfronthosting.co.za>
In-reply-to
Content
It sounds like doing this is fine (as Glyph suggests in the email thread) but it isn't magically going to solve all EINTR problems, just reduce the number of calls that could encounter them.

http://man7.org/linux/man-pages/man7/signal.7.html see the "Interruption of system calls and library functions" section.

Note that with this SA_RESTART flag set via siginterrupt you _may_ be making a trade off between being able to process python signal handlers during long reads or writes vs having to wait until the entire thing has finished.

given that, I'm ambivalent about this patch being worthy.

(re: Explicitly testing this, it is hard. for the broader Python test suite as a whole I've been pondering a regrtest mode that'll continually pester all of the test processes with signals to try and expose EINTR issues.  Low on my TODO ideas list, I haven't written code to do it.)
History
Date User Action Args
2013-12-01 19:40:24gregory.p.smithsetrecipients: + gregory.p.smith, gvanrossum, neologix
2013-12-01 19:40:24gregory.p.smithsetmessageid: <1385926824.0.0.338852467339.issue19850@psf.upfronthosting.co.za>
2013-12-01 19:40:23gregory.p.smithlinkissue19850 messages
2013-12-01 19:40:23gregory.p.smithcreate