[Python-Dev] xreadlines : readlines :: xrange : range

[Guido van Rossum]

[Me & Thomas in violent agreement that there's something weird about
the speed of getc_unlocked() vs. getc() on FreeBSD.]

I just realized what's the probable cause.  Read your timing post
again:

# BSDI:
# 
# (Python 2.0)
# while_readline        1.006  1.000
# 
# (CVS Python + getc_unlocked)
# while_readline        0.363  0.367

# FreeBSD:
# 
# (Standard CVS Python)
# while_readline        0.214  0.219
# 
# (CVS+getc_unlocked)
# while_readline        0.283  0.281

Standard CVS Python, as opposed to Python 2.0 as released, uses GNU
getline()!  So on FreeBSD, for this test case, GNU getline() is faster
than getc_unlocked().

So the question is, should I leave the GNU getline() code in?  I'm
inclined against it -- it's not that much faster, and on other
platform getc_unlocked() is faster.  Given that getc_unlocked() is a
standard (of some sort) and GNU getline() is, well, just that, I'd say
let's stick with getc_unlocked().

(Unfortunately, from a phone conversation I had last night with Tim,
there's not much hope of doing something there -- and that platform
sorely needs it!  The hacks that Tim reported earlier are definitely
not thread-safe.  While it's easy to come up with getc_unlocked() for
Windows, the locking operations used internally there by the /MT code
are not exported from MSVCRT.DLL, and that's crucial.)

--Guido van Rossum (home page: http://www.python.org/~guido/)