Re: patch@34561 VMS exec handling / cwd realpath fixes

Craig A. Berry wrote:
> On Thu, Oct 23, 2008 at 12:38 AM, John E. Malmberg <[email protected]> wrote:
>> Craig, Ken,
>>
>> Can you review this patch?
> 
> I'll take a look, though I can't promise to do so immediately.  At
> first glance the readdir part doesn't look right; there shouldn't need
> to be another completely different way of setting up the macros for
> one extension, but I don't know offhand why the existing macros would
> not be visible in File::Glob.

According to the comments in iperlsys.h, it is the place that the macros 
need to be in order to make the wrappers visible to Perl.

iperlsys.h had macros that where hiding the macros in vmsish.h.

The extra stuff in vmsish.h is for handling passing the thread context 
for some of the wrappers.

It would be nice to clean that up so that the wrappers for the C library 
do not need the thread context, so that they would be cleaner.

> There are really three separate patches here, the symlink one, the
> readdir one, and the Cwd one.  It would be helpful in the future if
> things like this could be submitted as three separate patches in three
> separate messages so they can be more easily reviewed, considered,
> tested, integrated, or (worst case) rolled back individually without
> affecting the unrelated parts.  For Cwd, the patches should be
> submitted here:
> 
> http://rt.cpan.org/Dist/Display.html?Name=PathTools
> 
> though for VMS-specific things I do appreciate a CC to vmsperl.
> 
> Thanks for your work, and for your patience while I try to find time
> to digest it.

Sorry about combining the patches, but I am trying to get back into 
getting the ODS-5/Unix/GNV mode, and the exec setup_cmddsc issue were 
blocking it, and I was trying to avoid taking three days to validate 
individual patches.

I thought that the readdir/bsd_glob issue might also be blocking me.

The symlink change forced the cwd change so as soon as the symlink 
change is commited, the cwd tests will start failing.

Changing cwd before the symlink change would also probably result in the 
cwd test failing, since the symlinks targets were being created in VMS 
syntax, so the realpath() routines would not have handled them as the 
test expects.

In UNIX compatibility mode, about 130 scripts out of 1424 are currently 
failing on VMS.

-John
[email protected]
Personal Opinion Only