Re: Threads - When?
Lourens Veen wrote:
The problem is not whether it will work or not, the problem is that
there is currently no way of writing a (guaranteed) portable C++
programme that uses threads. Currently, anything to do with threads
is undefined behaviour.
And IMO, that is the fault of the OS people and library writers, not
the language proper.
The C++ language semantics are specified in terms of an
abstract "virtual machine" to make the language specification
independent of any OS. If we want to implement a threading library,
then we have to specify its semantics in terms of that virtual
machine, just like with the rest of the language. And to be able to
define a mutex and say "No more than one thread shall own a mutex at
the same time", the virtual machine needs to support multithreading.
I would agree with this point of view under one condition: absolutely
no changes to C++ would be made. Note that I am not requesting that no
changes to C++ be made, I am trying to prove a point, which is...
Threading is an OS issue, not a language issue. It is OK to have
expectations of what the OS supports. I do not think it is ok that the
"power of the language" will make all better. In other words, I would
be entirely happy if the OS people provided a full set of primitives,
and the C++ built a portable library on top of those primitives.
So, the question is how to extend the C++ virtual machine to support
threads, and do it in a way that keeps it compatible with all the
mainstream operating systems in use today. Only when we have that can
we describe primitives and/or a standard library extension in terms
of it.
Write multi-threading applications first, *try* to build a multi-OS,
portable threading facade, and once you feel that you have a good idea
of the set of primitives that are required, go back to the OS people
(not C++ standards committee), and ask them to implement those
primitives on their OS's. Once that is done, normalize the threading
library, and then you are done.
I get the feeling that C++ programmers are, perhaps, expecting some new
keywords to be added to the language to make threading easier, or
whatever. Those keywords will end up being nothing more than library
calls in disguise. And most importantly, the _same_ attention to
detail will have to be exercise in the employment of those keywords as
would be required in the employment of thread-oriented library calls.
-Le Chaud Lapin-
--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]