Author skrah
Recipients Amaury.Forgeot.d'Arc, Jim.Jewett, Ramchandra Apte, amaury.forgeotdarc, benjamin.peterson, casevh, ced, eric.smith, eric.snow, haypo, jjconti, lemburg, mark.dickinson, pitrou, rhettinger, skrah
Date 2012-03-07.13:49:44
SpamBayes Score 3.01944e-09
Marked as misclassified No
Message-id <20120307134943.GA4928@sleipnir.bytereef.org>
In-reply-to <1331123950.3435.5.camel@localhost.localdomain>
Content
Antoine Pitrou <report@bugs.python.org> wrote:
> You could use Py_LOCAL_INLINE, but most compilers should inline small
> functions automatically, AFAIK.

At the time I wrote it I benchmarked everything. I'm pretty sure
that gcc did not inline larger functions like mpd_qresize_zero()
and mpd_word_digits() and even some smaller ones that are declared
ALWAYS_INLINE. Also, the static inline functions in the header
files are absolutely crucial for speed.

I recall that Mark initially said that in the Modules hierarchy
not every module would need to compile. Now, _decimal is already
tested with gcc, clang, icc, suncc, Visual Studio, and success
has been reported with xlc[1]. CompCert compiles libmpdec but not
Python.

_ctypes does not compile with icc and suncc. Unlike _ctypes, _decimal
has a fallback in the form of decimal.py. So, perhaps as an alternative
we could leave the inlines and wait for build failure reports?

[1] compilation success, one of the numerous AIX issues on
    bugs.python.org with loading the module was encountered IIRC.
History
Date User Action Args
2012-03-07 13:49:45skrahsetrecipients: + skrah, lemburg, rhettinger, amaury.forgeotdarc, mark.dickinson, pitrou, haypo, casevh, eric.smith, benjamin.peterson, jjconti, ced, Amaury.Forgeot.d'Arc, eric.snow, Ramchandra Apte, Jim.Jewett
2012-03-07 13:49:44skrahlinkissue7652 messages
2012-03-07 13:49:44skrahcreate