At 2:38 AM +0100 12/24/03, Odd Gripenstam wrote:

>When I build 5.8.2 from my home directory everything is OK except for the (still ?) 
>expected time zone problem:

Yes, unfortunately that one is still expected.

>$ sh def
> USER:[GRIPENSTAM.PERL-5_8_2]
>$ sh log user
>  "USER" = "$1$DGA25:[USER.]" (LNM$SYSTEM_TABLE)
>
>lib/vmsish...........................FAILED at test 25
>
>When I build from where we had planned to keep the sources I get other problems:

If by "keep the sources" you mean keep what is needed to build
extensions, you don't need the whole source tree, just what gets
copied by MMS INSTALL.

>The directory is $1$DGA20:[LAYERED.PERL.DISTRIBUTIONS.PERL-5_8_2] so
>I define a rooted logical name called PERLSRC per the readme file.
>This gives me an ACCVIO when building B.

An accvio with long pathnames is more than likely something just
recently fixed (since 5.8.2):

http://public.activestate.com/cgi-bin/perlbrowse?patch=21858

What's particularly puzzling is that since you use concealed as well
as terminal, it shouldn't be expanding the logical completely but
apparently is.

>$ mc [--]perl "-Px" SWITCHPX.AUX
>%DCL-W-TKNOVF, command element is too long - shorten

Once it truncates the command line, the rest of what happens is up
for grabs.  I don't see anything obvious about the test that would
make it more vulnerable to long filenames than other tests.  As long
as VMS has command buffers that are many times shorter than other
OS's, I think we will still see examples of this.

>Since I can build from my home directory this is mostly a FYI.

Thanks.  I may not have time to debug this in the near term, but it's
better to know about.  I'm glad you have a workaround.
-- 
________________________________________
Craig A. Berry
mailto:[EMAIL PROTECTED]

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

Reply via email to