[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Fix for back-shell child process deadlocks (ITS#2262)

On Thu, 9 Jan 2003, Kurt D. Zeilenga wrote:

> Basically, as POSIX did not define the semantics of pthreads
> and fork(), the behavior of fork() in pthread'ed applications
> is operating system/library system dependent.   Not only
> will your fix not solve the problem on all systems, it likely
> will cause some "working" systems to fail.

OK, now I understand. Thanks for clarifying this.

> I think the issue here is not really a matter of where I, you,
> or others might put their resources, because if it a portable
> generally-useful fix was developed it likely would be committed.

Although given the above, it sounds like there can never be a truly
portable fix unless we (a) don't use threads, or (b) don't use fork() -
neither of which sounds viable without a major rewrite. Either that or you
give up on complete portability, which is what I've chosen to do locally
but is clearly not acceptable for the project as a whole...

Simon Brady                             mailto:simon.brady@otago.ac.nz
ITS Technical Services
University of Otago, Dunedin, New Zealand