[Python-Dev] Compiling without multithreading support -- still useful?
Gregory P. Smith
greg at krypto.org
Wed Sep 6 16:06:04 EDT 2017
My take on platforms without thread support is that they should provide a
their own fake/green/virtual threading APIs. I don't know how practical
that thought actually is for things like web assembly but I'm with Antoine
here. The maintenance burden for --without-threads builds is a pain I'd
love to avoid.
On Wed, Sep 6, 2017 at 11:49 AM Ethan Smith <ethan at ethanhs.me> wrote:
> Certainly, I understand it can be burdensome. I suppose I can use 3.6
> branch for the initial port, so it shouldn't be an issue.
> On Wed, Sep 6, 2017 at 11:13 AM, Antoine Pitrou <solipsis at pitrou.net>
>> On Wed, 6 Sep 2017 10:50:11 -0700
>> Ethan Smith <ethan at ethanhs.me> wrote:
>> > I think this is useful as it can make porting easier. I am using it in
>> > attempts to cross compile CPython to WebAssembly (since WebAssembly in
>> > MVP does not support threading).
>> The problem is that the burden of maintenance falls on us (core CPython
>> developers), while none of us and probably 99.99% of our userbase have
>> absolutely no use for the "functionality".
>> Perhaps there's a simpler, cruder way to "support" threads-less
>> platforms. For example a Python/thread_nothreads.h where
>> PyThread_start_new_thread() would always fail (and with trivial
>> implementations of locks and TLS keys). But I'm not sure how much it
>> would help those porting attempts.
>> Python-Dev mailing list
>> Python-Dev at python.org
> Python-Dev mailing list
> Python-Dev at python.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Python-Dev