[Info-vax] Perl (for OpenSSL) v. me

[Craig A. Berry]

On 2/9/16 2:25 AM, richard at levitte.org wrote:
> Den söndag 7 februari 2016 kl. 08:58:58 UTC+1 skrev Steven Schweda:
>> M. Levitte suggests that Perl 5.10.1 worked for someone on VMS.
>

> Not quite.  I believe you refer to this part of my announcement on
> openssl-dev:
>
>> - Perl!  Reports tell me that version 5.10.1 works fine but might
>> need
>
> That was perl in general, not perl on VMS specifically.  The report
> is based on what versions of core modules came with different
> versions of perl, and someone (I can't remember who for the moment)
> helped me determine that 5.10.1 had good enough compatibility.  Of
> course, that person is based on Unix.

5.10.1 was a good release and an important one given that it broke the
logjam and got releases going again after a few years without one. But
by my count there have been 25 stable releases since then (current is
5.22.1) and I can't think of a good reason to be using a release that
old anymore.

> But anyhow, this discussion tells me I should avoid realpath() like
> the plague, at least on VMS.  I'll see what I can figure out (I think
> rel2abs() from File::Spec might do the trick).

I agree. The more I think about it, the more it seems to me that using
realpath() and also using abs2rel() are at odds with each other on any
platform. realpath() is supposed to peel back on the convenience layers
(symlinks, concealed logicals, etc.) and abs2rel(), at least when called
without its optional base argument, is supposed to give you a path
relative to the current working directory of the current user session,
which is expressed in terms of those convenience layers.