Message136333
| Author |
neologix |
| Recipients |
alexey-smirnov, amaury.forgeotdarc, neologix, socketpair, vstinner |
| Date |
2011年05月19日.20:34:19 |
| SpamBayes Score |
3.0514748e-08 |
| Marked as misclassified |
No |
| Message-id |
<BANLkTi=mHeNQeixwPEK08Y5QJ1BZ8DhF+g@mail.gmail.com> |
| In-reply-to |
<1305832532.67.0.252255366596.issue12105@psf.upfronthosting.co.za> |
| Content |
> Using spawn_python() to check that os.O_CLOEXEC flag is correctly set seems
> overkill. Why not just testing fcntl.fcntl(f.fileno(), fcntl.F_GETFL) &
> FD_CLOEXEC)?
Because I couldn't find a place where the CLOEXEC flag was fully
tested (I mean, checking in the child process that the FD was
correctly closed), so I took the opportunity to test it thoroughly
here.
But you're right it's maybe a little bit overkill, so here's a patch
using just F_GETFL. Pick up whichever you like.
>> Note that I'm not sure that adding this flag to built-in open()
>> is necessarily a good idea
>
> I agree.
>
OK.
> open() documentation may explain the os.fdopen(os.open()) "trick" to use
> low-level options like O_SYNC or O_CLOEXEC.
>
Why not, but I leave it to someone more comfortable with documentation
than me :-) |
|