Re: [Module::Build] Re: Module::Build status on VMS

[Craig A. B. <cra...@ma...>]

At 9:31 PM -0500 1/26/06, John E. Malmberg wrote:
>Craig A. Berry wrote:
>>
>>So now we're getting to the problem that was Yitzchak's first theory,
>>namely that it's a case problem, more specifically, the case-leveled
>>names returned from File::Find do not match the case-preserved names
>>in the hash, so we delete the files because they are not recognized.
>>One way to deal with this would be to acquire or build a
>>case-tolerant tied hash and use one of those instead of an ordinary
>>hash.  Basically when you look up a case-preserved name in the hash
>>and it doesn't match, it will then look up a case-leveled version for
>>you.  Schwern suggested this solution a long time ago for
>>ExtUtils::Manifest::manifind() and it's still needed there, so
>>perhaps we could put such a package in a place where both MM and MB
>>can use it.  I think there are a couple of implementations on CPAN .
>
>Try
>
>$DEFINE DECC$EFS_CASE_PRESERVE ENABLE
>
>And blead will preserve the case of files on ODS-5 volumes. 

Assuming previous activities, such as unpacking the source archive or
applying patches to it, have not already leveled the case.  This
could be a way to set aside the case issue and see what else comes
up, but the case issue still needs to be solved.

>On ODS-2 volumes they will be uppercased.

Which is a bit of a problem because it's the exact opposite of what
the default behavior was when ODS-2 was the only filesystem the C
run-time knew about.

>
>Note some of the lib and external modules may still be assuming that VMS means all lowercase filenames because this feature is still not implemented fully.

Right.  And what I'm proposing is a way to start getting some of the
key infrastructure ready to handle it.
-- 
________________________________________
Craig A. Berry
mailto:cra...@ma...

"... getting out of a sonnet is much more
 difficult than getting in."
                 Brad Leithauser