Re^4: HPUX corruption of file handle after fork

Aha - so it's not the perl version but it's probably perl on HPUX.

I'm not into this enough to look at perl source code, but I find it hard to imagine it's got something special about the way it calls close() for HPUX.

H'm. To me it means I have to be careful with explicitly opened files when calling fork(). perl open() is probably safe.

If you find out more do let me know, but I'm willing to use fork() now, just I know what to expect.

Thanks - Abe

Comment on Re^4: HPUX corruption of file handle after fork

Replies are listed 'Best First'.
Re^5: HPUX corruption of file handle after fork by sgt (Deacon) on Dec 10, 2007 at 22:50 UTC
There are lots of issues at hand. At the Unx/POSIX level exit() does lots of cleaning. To be on the safe side it is usually better to _exit() in the child. IO buffering can get in the way as exit() will usually flush the streams before closing them. perl will usually show its unx heritage (except in some places and/or platforms in recent years where it was considered better to "fix" that un*x behaviour). Recently while translating C code for a system(3) with time-out I needed low-level IO and POSIX::_exit() to be close to the old behaviour. Even at the perl level don't forget that exit() does call END blocks ... 5.8.0 was very buggy, so it would be better to use another version. cheers hth --stephan	[reply]
Re^6: HPUX corruption of file handle after fork by Abe (Acolyte) on Jan 31, 2008 at 16:57 UTC
Thanks sgt - sorry I didn't notice for a bit. I'd solved it and didn't check the thread again. I was a chicken and just stuck to STDIN, though I'd understood what had gone wrong. Regards, Abe	[reply]