Re: [Fwd: Re: Perl 5.12.2-RC1 is now available]

On Sep 1, 2010, at 12:55 PM, Eric Robertson wrote:

> A few days ago I downloaded the perl-5.12.1.tar.gz tarball from CPAN  
> and built it on my Alphaserver ES40 running OpenVMS V8.3 and DECC  
> V7.3-009. I have attached a file containing selected excerpts from  
> that build. My build of the previous 5.12.1 release produced fewer  
> test failures (17) than did than Martin Vorländer's reported build  
> of 5.12.2 RC1 (37). So, judging solely by the number of failed  
> tests, it looks like 5.12.2 RC1 might have broken more things than  
> it fixed.

That's extremely unlikely -- it's a pretty modest and conservative  
release.  Of course if there is something seriously wrong, I want to  
know about it.

For good or for ill a lot of it comes down to environmental  
differences: note Martin and me getting 37 vs. 0 test failures, but  
both on OpenVMS I64 v8.3-1H1 with exactly the same compiler version.   
Some of the tests seem to be very sensitive to the environment.   
Different definitions of DECC$* logicals and others, different quotas  
and protections, different volume settings (special files on or off,  
write-behind caching on or off), etc.

It would be nice to get a handle on all of these differences and make  
the tests immune to them, but that's not something I can do alone.  So  
I do appreciate reports, especially those that come with analysis of  
exactly what went wrong and why.

> Also, there are some warnings that I have included in the attached  
> excerpt file. I was not sure if these warnings were significant  
> (this is the first time that I have built perl from scratch on  
> OpenVMS). So you can ignore them if experience tells you that they  
> are not significant.
>

I think a lot of the warnings are not VMS-specific, and are probably  
not optimal but aren't considered a high priority.  The long symbol  
truncations are VMS-specific and are there to warn anyone linking an  
extension against an external library that there may be trouble.

________________________________________
Craig A. Berry
mailto:[email protected]

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