Message253247
> It will be nice, if someone comment that hacks in source code
I don't understand why you keep calling this a hack. It's part of open() contract, and I'm quite sure that it was a deliberate choice to declare O_TMPFILE as O_DIRECTY|new_bit. See for example this comment:
https://lwn.net/Articles/560834/
I wrote a patch to explain that it's fine to call open() with O_TMPFILE on old kernels to check if the flag is supported: see attached patch. |
|
Date |
User |
Action |
Args |
2015-10-20 19:39:50 | vstinner | set | recipients:
+ vstinner, georg.brandl, ncoghlan, pitrou, Arfrever, neologix, socketpair, python-dev, serhiy.storchaka, josh.r |
2015-10-20 19:39:50 | vstinner | set | messageid: <1445369990.52.0.181711076719.issue21515@psf.upfronthosting.co.za> |
2015-10-20 19:39:50 | vstinner | link | issue21515 messages |
2015-10-20 19:39:50 | vstinner | create | |
|