On Thursday February 12 2015 11:28:15 Joshua Root wrote:
>Even if you don't deliberately do anything between the fork and exec,
>it's possible for things to happen because of signals and the like. This
>is one reason why Apple recommends using posix_spawn.
Do you have a valid link to that? The one currently in the comment Ian turned
up ("See "CoreFoundation and fork()" at
http://developer.apple.com/releasenotes/CoreFoundation/CoreFoundation.html) is
stale, the page no longer evokes fork at all).
I'm beginning to think that we should probably do a survey of the cases in
which fork() is used and see to what extent a wrapper class or helper function
couldn't be written in which we could use the Apple-sanctioned way so that the
issue is dealt with once and for all. If Qt doesn't already have something like
that it may be a useful addition, too.
Come to think of it, MS Windows doesn't have fork() AFAIK. How has that been
dealt with?
NB: if a wrapper class or helper function is feasible, it should also be
possible to begin developing it in a portable way on the platform we currently
have, i.e. KDE4 (evidently not necessarily with the goal of deploying it
throughout).
R.
_______________________________________________
macports-dev mailing list
[email protected]
https://lists.macosforge.org/mailman/listinfo/macports-dev
