Re: regex is much slower in pypy than cpython

Michał Górny <[email protected]> Wed, 03 Sep 2025 06:15:40 +0200
Newsgroups gmane.comp.python.pypy
Organization Gentoo
Message-ID <[email protected]>
--===============5082396755538473672==
Content-Type: multipart/signed; micalg="pgp-sha512";
	protocol="application/pgp-signature"; boundary="=-6F7GmOq+wVQ8rsAdJ877"


--=-6F7GmOq+wVQ8rsAdJ877
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Wed, 2025-09-03 at 10:16 +0800, Li Li via pypy-dev wrote:
> I want to use pypy to speed up my application, it's fast for standard
> library such as dict/list. I need to use a third party library regex(
> https://pypi.org/project/regex/). pypy is much slower with this library.
> for full codes, see https://github.com/mrabarnett/mrab-regex/issues/586
>=20

The regex library doesn't work correctly under PyPy.  The author is
using internal CPython implementation details that don't match PyPy,
meaning it's not only slow but also incorrect.  We have reported this
before, suggesting how to achieve the same thing using the correct,
public API, but the author was opposed to it.

My suggestion would be not to use that library at all.  Some of its
functionality is provided on top of stdlib by "backrefs" package, maybe
that'll help.


--=20
Best regards,
Micha=C5=82 G=C3=B3rny

--=-6F7GmOq+wVQ8rsAdJ877
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: This is a digitally signed message part

-----BEGIN PGP SIGNATURE-----

iQFGBAABCgAwFiEEx2qEUJQJjSjMiybFY5ra4jKeJA4FAmi3wOwSHG1nb3JueUBn
ZW50b28ub3JnAAoJEGOa2uIyniQOErkIAJ2dcmwHbr2UQHokFmUKTL6WGlphYX2z
VQjk/Q9WC/blo3fyHq7Z1u/Xaj6ZbR+Rgd2VtRjURRwW78Y+leGEWtUDBz3rCegm
RC3x9XWqJy/iDjslQocxvnaDjNnFpWYpzgAYiag9J6Zpkb89K3DVrympvq3py8h/
/tXqjji0WwCaIWTnoqm8eZ2Wbxo6I/q972dglN0OEB2QvKcSQxdf12+KTaBWL86A
KJT/mdEwXjbk7ioiuEIs4s6yT7r9sJtGxTmpmlL/P7z1A8VVvlRKFhwPzTyeja9p
KF2EMt+EW2m+dVSC4lSl+WhQYNxxifHJmRFZ2ig1GWjmEje3LXdpEUY=
=Ghti
-----END PGP SIGNATURE-----

--=-6F7GmOq+wVQ8rsAdJ877--

--===============5082396755538473672==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
pypy-dev mailing list -- [email protected]
To unsubscribe send an email to [email protected]
https://mail.python.org/mailman3//lists/pypy-dev.python.org
Member address: [email protected]

--===============5082396755538473672==--