I am moving home this week and do not have time to review. But I want to notice about the difference between --enable-X and --with-X. See https://autotools.io/autoconf/arguments.html
Since isolated subinterpreter is not external dependency, "--enable" should be used here. (I know we already abuse "--with-lto".) Regards, On Thu, May 7, 2020 at 2:50 AM Victor Stinner <vstin...@python.org> wrote: > Hi, > > tl; dr I'm asking for your permission to merge the following PR :-) > > https://github.com/python/cpython/pull/19958 > > > In bpo-40514, I added a new > --with-experimental-isolated-subinterpreters configuration option. I > chose to use a very long option name and to not document it on > purpose: prevent users to use it "by mistake", without understanding > its purpose. The option is related to "per-interpreter GIL": > > https://bugs.python.org/issue40512 > > It's a practical solution to be able to experiment quickly > per-interpreter GIL without having to fix all issues at once. For > example, it disables Unicode interned strings which is unsafe with > multiple interpreters running in parallel. > > I added this #ifdef to encourage other core developers work on this > project, and let early adopters test this experimental feature to give > us their feedback. > > I hope that soon we will discover all places which need to be fixed, > so it will help to better estimate how much work is needed to finish > the implementation of per-interpreter GIL. > > Currently, the special build changes: > > * Per-interpreter GIL > * Store the current Python thread state in a TLS > * Disable dict, frame, tuple and list free list > * Disable type method cache > * Disable pymalloc: force usage of libc malloc > * Disable the GC in subinterpreters > * _xxsubinterpreters.run_string() releases the GIL > > I consider that's a reasonable small number of changes to get a cool > feature (per-interpreter GIL), compared to the same of the CPython > code base (637K lines). > > -- > > All these "#ifdef EXPERIMENTAL_ISOLATED_SUBINTERPRETERS" are temporary > workarounds until a proper fix is designed. For example, some caches > should be made "per-interpreter". > > Most changes are easy to write, but some other changes are non > trivial. For example, I modified _PyThreadState_GET() and > _PyThreadState_Swap() to use a Thread Local Storage (TLS) to get and > set the current Python thread state. > > Currently, "#ifdef EXPERIMENTAL_ISOLATED_SUBINTERPRETERS" are not > "visible" to users: it is only used in .c files and a few internal > header files. But the following PR modify the public Include/object.h > header file to make PyObject.ob_refcnt atomic: > > https://github.com/python/cpython/pull/19958 > > I would like to ensure that you are ok to put a few more temporary > "#ifdef EXPERIMENTAL_ISOLATED_SUBINTERPRETERS" in CPython to speedup > the development of subinterpreters running in parallel > (per-interpreter GIL), or if you consider that it has gone too far and > a Git fork would be better. > > Victor > -- > Night gathers, and now my watch begins. It shall not end until my death. > _______________________________________________ > python-committers mailing list -- python-committers@python.org > To unsubscribe send an email to python-committers-le...@python.org > https://mail.python.org/mailman3/lists/python-committers.python.org/ > Message archived at > https://mail.python.org/archives/list/python-committers@python.org/message/2WHQBHFK7UQ7OYAJV5S2QCINODVVKGJY/ > Code of Conduct: https://www.python.org/psf/codeofconduct/ > -- Inada Naoki <songofaca...@gmail.com>
_______________________________________________ python-committers mailing list -- python-committers@python.org To unsubscribe send an email to python-committers-le...@python.org https://mail.python.org/mailman3/lists/python-committers.python.org/ Message archived at https://mail.python.org/archives/list/python-committers@python.org/message/4F67S3G7ME64G2R73326TWOCFX5XNNMB/ Code of Conduct: https://www.python.org/psf/codeofconduct/
