[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: perl ithreads support: why hardcode at configure time ?
From: |
Mike Frysinger |
Subject: |
Re: perl ithreads support: why hardcode at configure time ? |
Date: |
Fri, 11 Jan 2013 12:11:05 -0500 |
User-agent: |
KMail/1.13.7 (Linux/3.7.1; KDE/4.6.5; x86_64; ; ) |
On Friday 11 January 2013 04:08:26 Stefano Lattarini wrote:
> On 01/11/2013 05:07 AM, Mike Frysinger wrote:
> > i can't imagine this is a big runtime penalty, so why does configure
> > check for the perl's thread settings and then hardcode that in the
> > generated automake ?
>
> I don't know, I wasn't around when the change was introduced. You'll have
> to dig out the archives; if we don't find any compelling reason for having
> the check at configure time rather than runtime, I agree your change is a
> nice simplification ...
i couldn't find references to ithreads in the automake or automake-patches
archives
> > --- a/configure.ac
> > +++ b/configure.ac
> > @@ -77,32 +77,6 @@ installed, select the one Automake should use using
> > ./configure PERL=/path/to/perl])
> > }
> > -# We require ithreads support, and version 5.7.2 for CLONE.
>
> Here, the comments say (and the code agree) that we should support
> ithreads only from perl 5.7.2 onwards ...
i thought that meant something else. oh well.
> > +use Config;
> > +our $perl_threads = $Config{useithreads};
>
> ... but there is no such check here. And yes, so far we still support
> perl 5.6 so far (albeit we might want to start requiring 5.8 in Automake
> 1.14, but that's for another thread).
i'm not a perl hacker, so i don't know how to do that. but a grep shows that
automake is already doing that sort of thing in the tests, so i just copied
that:
+our $perl_threads = 0;
+if (eval { require 5.007_002; }) # for CLONE support
+ {
+ use Config;
+ $perl_threads = $Config{useithreads};
+ }
> Also, might I ask you to format your patches with "git am"? That will make
> it super-easy for me to apply them.
i'm familiar with git. i expected more push back on this, so i didn't write a
proper commit as this was a RFC with a PoC patch inline. no point in wasting
time doing a write up (or the god awful time waste known as the GNU ChangeLog)
if the patch is going to get rejected.
-mike
signature.asc
Description: This is a digitally signed message part.