Nicholas Clark wrote:
> Personally I'm not going to start work on such a thing, as I'm not really
> interested, and have other things I need (working h2xs constant()
> generator being a new one) and need to do (*****y NaNs) but I'm not stopping
> anyone else dedicating tuits to the task if they feel that it is good.
> Additionally I'm advocating not discouraging others who feel strongly on the
> matter attempting to *do* something about it. [as distinct from others
> *saying* things in an attempt to encourage third parties to do something about
> it]
I'd be delighted to see a correctly implemented, safe threads implementation
in perl. I would be extremely unhappy to see something that we only think
works, but actually doesn't, or that only works on a small subset of
platforms. Pointing out the localtime() problem was an easy target - if
that was missed, how many of the difficult problems have been missed as
well?
I wonder what proportion of the Ithreads code has really been heavily
stressed on significantly-sized MP boxes. As it is mainly used in a Win32
environment, I suspect this hasn't yet happened. I think it is tempting but
dangerous to make statements about Ithreads based mainly on uniprocessor
Win32 that will prove to be false on different platforms.
Alan Burlison
