Messages in this thread | | | Date | Tue, 28 Jul 2009 09:52:38 -0700 (PDT) | From | Linus Torvalds <> | Subject | Re: [PATCH] kdesu broken |
| |
On Tue, 28 Jul 2009, Linus Torvalds wrote: > > The correct expectation is that the select() (or read()) should have > returned any data that it saw _before_ it returns EINTR.
Put another way: our pty code is simply _buggy_ if it returns EINTR when there is actually data pending on a pty.
Yes, it's "buggy" only in a QoI sense - I'm sure that if you read POSIX and SuS like a language lawyer, there is absolutely zero that says that it can't return EINTR at any random time, or that the SIGCHLD should be ordered wrt the other side of the pty having done a write.
But we don't do language-lawyering based on standards that inevitably never really delve into all the nitty-gritty details. We are simply better than that. Leave the language-lawyering to the people who can't do things well, and then whine about their crap being "technically correct".
Linus
| |