Ionitor has asked for the wisdom of the Perl Monks concerning the following question:
This works fine in Unix Perls, but in 5.6.x, the Windows fork emulation had... issues. This was due to a problem in the new ithreads model, as the regex engine caused some race conditions. This would result in programs that worked fine most of the time, but sometimes certain inputs would cause perl to crash. Sometimes, this crash could be eliminated just by switching two lines of code around. In other words, it wasn't terribly predictable.
So I now find myself getting ready to write another program that will successively filter data; however, it needs to run under Windows. I have some other ideas how to avoid forking, but they all make the program much more complicated than I would like. So here's my question (finally): has anyone worked with fork emulation since the regex issues were resolved (in one of the 5.7.x's)? If so, have you had any problems you've attributed to other fork emulation problems?
If you haven't worked with fork emulation (because you don't code for Windows or don't fork there), has anyone used ithreads enough to comment on their stability? Fork emulation is based on ithreads, so if they are stable, I would guess fork is almost as good.
I realize that fork emulation/ithreads are definitely not perfect and that there are still bugs and limitations listed in the docs--particularly the possibility of bugs in XS modules. However, none of those are issues that I'm worried about at the moment. I just want to know if there are still misc. bugs that will cause me problems.
|
|---|
| Replies are listed 'Best First'. | |
|---|---|
|
Re: Perl 5.8.0 fork emulation
by Anonymous Monk on Jul 23, 2002 at 03:30 UTC |