Comments on: Code review: libreplace http://rusty.ozlabs.org/?p=61 Stealing From Smart People Thu, 19 Aug 2010 18:12:38 +0000 http://wordpress.org/?v=2.9.2 hourly 1 By: Adam Kennedy http://rusty.ozlabs.org/?p=61&cpage=1#comment-169 Adam Kennedy Tue, 16 Feb 2010 04:06:37 +0000 http://rusty.ozlabs.org/?p=61#comment-169 Now that I think about it more, some kind of direct CCAN -> CPAN wrapper mechanism would be a great thing to work on at the unconference. If you can make it work, you'd also get free access to CPAN Testers for the entire of CCAN... Now that I think about it more, some kind of direct CCAN -> CPAN wrapper mechanism would be a great thing to work on at the unconference.

If you can make it work, you’d also get free access to CPAN Testers for the entire of CCAN…

]]> By: Adam Kennedy http://rusty.ozlabs.org/?p=61&cpage=1#comment-168 Adam Kennedy Tue, 16 Feb 2010 04:03:46 +0000 http://rusty.ozlabs.org/?p=61#comment-168 It should be pretty easy to bootstrap this up into a Perl distro that can be pushed out to CPAN Testers. Maybe something like an Inline::C wrapper over the functions to allow a bunch of Perl test scripts to be written would be enough. If you can link the two together CCAN -> CPAN in that way, you could get access to some of the more interesting "strange" platforms, like HPUX/S390/VMS/EBCDIC-things etc. It should be pretty easy to bootstrap this up into a Perl distro that can be pushed out to CPAN Testers.

Maybe something like an Inline::C wrapper over the functions to allow a bunch of Perl test scripts to be written would be enough.

If you can link the two together CCAN -> CPAN in that way, you could get access to some of the more interesting “strange” platforms, like HPUX/S390/VMS/EBCDIC-things etc.

]]> By: rusty http://rusty.ozlabs.org/?p=61&cpage=1#comment-160 rusty Sat, 13 Feb 2010 01:36:16 +0000 http://rusty.ozlabs.org/?p=61#comment-160 Ah, good point. PATH_MAX, of course, <a href="http://insanecoding.blogspot.com/2007/11/pathmax-simply-isnt.html" rel="nofollow">isn't</a>. Ah, good point. PATH_MAX, of course, isn’t.

]]> By: rusty http://rusty.ozlabs.org/?p=61&cpage=1#comment-159 rusty Sat, 13 Feb 2010 01:25:54 +0000 http://rusty.ozlabs.org/?p=61#comment-159 I had never heard of gnulib! Who knew? I like the source distribution, dislike the coding style and poor documentation (I'd like more hint as to why you'd want a function, such as an example). I grabbed a random module to review, and it was a little flawed: <a href="http://git.sv.gnu.org/gitweb/?p=gnulib.git;a=blob_plain;f=lib/nanosleep.c" rel="nofollow">nanosleep.c</a>. The signal handling is fubar: it mugs SIGCONT without restoring it (which seems outright weird, assume it's a special case someone needed), and returns 1 if it happens (not -1); any other signal which breaks select() out is silently ignored. Looks like it's "what we needed" rather than a really generic replacement. But other than "you should read the code before importing it", I think it's an excellent idea. As you say, at least someone else maintains it! I had never heard of gnulib! Who knew? I like the source distribution, dislike the coding style and poor documentation (I’d like more hint as to why you’d want a function, such as an example).

I grabbed a random module to review, and it was a little flawed: nanosleep.c. The signal handling is fubar: it mugs SIGCONT without restoring it (which seems outright weird, assume it’s a special case someone needed), and returns 1 if it happens (not -1); any other signal which breaks select() out is silently ignored.

Looks like it’s “what we needed” rather than a really generic replacement. But other than “you should read the code before importing it”, I think it’s an excellent idea.

As you say, at least someone else maintains it!

]]> By: Stewart Smith http://rusty.ozlabs.org/?p=61&cpage=1#comment-158 Stewart Smith Fri, 12 Feb 2010 23:53:57 +0000 http://rusty.ozlabs.org/?p=61#comment-158 For drizzle, we've been using gnulib (http://www.gnu.org/software/gnulib/) and it seems to be working pretty well. At least for the platforms where things are strange (Solaris and OSX). it does have the benefit of being maintained by someone else :) For drizzle, we’ve been using gnulib (http://www.gnu.org/software/gnulib/) and it seems to be working pretty well.

At least for the platforms where things are strange (Solaris and OSX).

it does have the benefit of being maintained by someone else :)

]]> By: Alex http://rusty.ozlabs.org/?p=61&cpage=1#comment-156 Alex Fri, 12 Feb 2010 11:51:03 +0000 http://rusty.ozlabs.org/?p=61#comment-156 Isn't PATH_MAX a relic anyway (along with ARG_MAX)? I thought in modern systems you should ask the OS what the limits are rather than rely on #defines. Isn’t PATH_MAX a relic anyway (along with ARG_MAX)? I thought in modern systems you should ask the OS what the limits are rather than rely on #defines.

]]> By: Jonathan Nieder http://rusty.ozlabs.org/?p=61&cpage=1#comment-155 Jonathan Nieder Fri, 12 Feb 2010 09:32:40 +0000 http://rusty.ozlabs.org/?p=61#comment-155 How does this compare to gnulib? Are there strong reasons to use each rather than the other? How does this compare to gnulib? Are there strong reasons to use each rather than the other?

]]>