[K42-discussion] Another problem with asynchrony

[Bryan S Rosenburg]

"Paul H. Hargrove" <PHHargrove at lbl.gov> wrote on 10/11/2005 05:30:54 PM:
> ISO C says
> 
> If a signal handler is called, the raise function shall not return until 
after
> the signal handler does.

That sounds like a requirement on signals generated by the program itself. 
 It's still not clear to me that it applies to a SIGPIPE generated in the 
operating system.  But I'm just splitting hairs.  We usually want to mimic 
Linux behavior whether or not that behavior is required by spec.

> 
> I am not imagining that write will "care" that it is talking to a pipe,
> just that the return value is EPIPE:
> 
> if (result == EPIPE) raise(SIGPIPE);

This is still a completely superfluous check if the write is to a regular 
file.  Yes, it's just a cycle or two at most, but once it's in there 
people will be tempted to add other things that only apply in special 
scenarios.  As far as I'm concerned, the whole SIGPIPE thing is an ugly 
wart, and I'd like to see its implementation confined to the server that 
implements pipes.

- Bryan

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://ozlabs.org/pipermail/k42-discussion/attachments/20051012/48862e23/attachment.htm