Author strombrg
Recipients Christophe Simonis, Garen, amaury.forgeotdarc, arekm, devurandom, doko, eric.araujo, georg.brandl, jreese, lars.gustaebel, leonov, loewis, nicdumz, nikratio, ockham-razor, pitrou, proyvind, rcoyner, strombrg, thedjatclubrock, vstinner, ysj.ray
Date 2010-11-26.04:51:50
SpamBayes Score 4.10592e-07
Marked as misclassified No
Message-id <1290747112.44.0.514111189261.issue6715@psf.upfronthosting.co.za>
In-reply-to
Content
I agree that ctypes is a bit more brittle - both ctypes and c extension modules can yield segfaults, but at least the c extension module is likely to give an error or warning when you rebuild it.

However, I'm getting the impression that:
1) In pypy, the C extension module layer isn't a complete emulation
2) In pypy, the C extension module is a bit disfavored compared to ctypes
3) In pypy, there's a performance expense for CPyExt compared to ctypes
3) (less important) In pypy, the C extension module layer is still too young to do much with at this time

I've been wondering if maybe there should be a way to check ctypes use against .h's at build time.  That way, the brittleness should be about the same.
History
Date User Action Args
2010-11-26 04:51:52strombrgsetrecipients: + strombrg, loewis, georg.brandl, doko, amaury.forgeotdarc, arekm, lars.gustaebel, pitrou, vstinner, nicdumz, eric.araujo, Christophe Simonis, rcoyner, proyvind, nikratio, leonov, devurandom, Garen, ysj.ray, thedjatclubrock, ockham-razor, jreese
2010-11-26 04:51:52strombrgsetmessageid: <1290747112.44.0.514111189261.issue6715@psf.upfronthosting.co.za>
2010-11-26 04:51:50strombrglinkissue6715 messages
2010-11-26 04:51:50strombrgcreate