jacob navia
7/4/2011 5:21:00 PM
Le 04/07/11 18:01, James Kuyper a écrit :
> On 07/04/2011 11:29 AM, Noob wrote:
>> Malcolm McLean wrote:
>>
>>> It's largely fixed. There was an attempt to update the language to
>>> include features like variable-sized arrays [i.e. C99], however it
>>> failed to be widely accepted. The user community preferred the
>>> language as it was.
> The multi-threading features seem to be on track for approval, so his
> comments imply that those changes will be just as unpopular as the ones
> made in C99. Since support for multi-threaded code seems to be one of
> the most popular requests in this forum, that suggests a judgment the
> multi-threaded support in C1X is so poorly designed that it won't
> actually be used. Is it? I can't tell, I've insufficient experience with
> such things to judge.
That will be a catastrophe.
There are mainly two competing standards that take 100% of the multi-
threaded code in C:
(1) POSIX pthreads
(2) Windows threads
The idea that people will REWRITE their threading code to please a
standard that isn't debugged, and has (at the start) ZERO support
is completely unconnected with software construction realities.
The features that were added to C99 didn't get wide support because
they weren't really essential but they were completely easy to
implement (and for many) GNU had already broken ground with them.
Implementing multi threading support however is completely different.
This needs a LOT of care to implement, and compiler vendors
will hesitate to implement something nobody asked them to do.
You say:
> Since support for multi-threaded code seems to be one of
> the most popular requests in this forum...
I do not remember ANYBODY asking for multi-threading support in
this forum for the pas 10 years or so, in any case as my
memory serves
If you search with google you will find a thread of about 2005
when somebody wrote here he was writing a multi-threaded TCP
server and he got the usual answer that multi-threading is system-
specific and should go somewhere else.
And you tell us "the most popular request"...
It is obvious that you want t support the committee, and maybe it is
right to do so. Bending the truth is not a good strategy however.
In this forum (and in comp.std.c) NOBODY has asked for that. The
committee decided to include that because Mr Plaugher decided that
he wanted that in the standard, not because in the user community
somebody asked for that.
The first versions of the specs were just a COPY AND PASTE from the
documentation of Plaugher's multi-thread library.
Maybe that has changed, I did not follow that since I consider that
the language can't do a THIRD specification that will ADD TO THE
CONFUSION of already two competing threading models.
I see that as completely ridiculous