Re: Relative speed of globbing and regex implementations

2007-12-06 Thread Jim Meyering
"James Youngman" <[EMAIL PROTECTED]> wrote: > I notice that the gnulib fnmatch implementation is slower than the > regex implementation, at least in those cases where a pattern can be > compiled once and offered thousands of times. Very roughly, I'd guess > that the fnmatch implementation takes a

Relative speed of globbing and regex implementations

2007-12-06 Thread James Youngman
I notice that the gnulib fnmatch implementation is slower than the regex implementation, at least in those cases where a pattern can be compiled once and offered thousands of times. Very roughly, I'd guess that the fnmatch implementation takes about 1.4x the time that the regex implementation take

Support versions of autoconf prior to 2.59c.

2007-12-06 Thread Jim Meyering
I discovered that gnulib-tool requires autoconf-2.59c or newer for autoconf's m4_foreach_w macro. This change lets gnulib-tool work also with autoconf-2.59. This was necessary for libvirt, when running their ./autogen.sh script on a RHEL5 system, which has autoconf-2.59: http://thread.gmane.org

Avoid a configure-time syntax error in gl_FUNC_ACL.

2007-12-06 Thread Jim Meyering
Using the latest gnulib code in coreutils, I got a configure-time syntax error because the cache variable was not defined in the else branch. It's very tempting to hoist the seemingly duplicate AC_CHECK_FUNCS call, but you can't, since it to be run after the library check. I've just pushed this cha

Re: base64 build problem on DragonFly 1.8.0

2007-12-06 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Martin Lambers on 12/5/2007 1:29 PM: >> ./stdio.h:372: error: syntax error before "getdelim" It sounds like ssize_t is not defined at this point. Yet, at the top of stdio.in.h, there is this chunk: #if (@GNULIB_FSEEKO@ && @REPLACE_FSEE