[Clfs-dev] issues with current svn x86_64 pure 64

[Martin Ward]

Ken Moffat wrote:
> On Sun, Jan 04, 2009 at 02:52:05PM -0500, Joe Ciccone wrote:
>   
>>> i'm going from memory here, yep otherwise nothing else would have
>>> built correctly, i'll have another look later in the week, i used make
>>> perl utilities ext/Errno/pm_to_blib to get errno which is needed for
>>> the coreutils testsuite
>>>       
>> Actually, this is really one of the first things you're building with
>> that compiler (temp-system). The entire chapter 6 (temp-system) build is
>> built using the cross-compiler built in chapter 5 (cross-tools).
>>     
>  In this context,  I read "otherwise nothing else would have built
> correctly" as referring to the final system.  And if Martin had not
> already replied, I would have said the same as he did here.
>
>  I haven't run the system yet, but sanity checks during the build,
> and ldd on random files, all look good.  The problem is purely
> within the temporary perl.  Now that I've built the final system, I
> can no longer replicate the problem.
>
> ĸen
>   
Correct, running on a full kde-3.5.10 desktop built from CLFS SVN 
11-11-08 (or round about), i still think the problem is with the temp 
perl (chap 10.4).

Looking at Errno_pm.PL in perl-5.10.0/ext/Errno
it tries to find the location of errno.h so it then can extract the 
system specific error numbers, however our errno.h is still sitting in 
tools/include, we haven't put the anything in usr/include yet
the test is around line 148/150
> # Some Linuxes have weird errno.hs which generate
>     # no #file or #line directives
>     my $linux_errno_h = -e '/usr/include/errno.h' ?
>         '/usr/include/errno.h' : '/usr/local/include/errno.h';
>     $file{$linux_errno_h} = 1;
then continues

when you've built the final system errno.h is in it's usual 
place(/usr/include) and the test passes

Martin