At 02:46 PM 5/10/00 -0700, [EMAIL PROTECTED] wrote:
>Dan Sugalski wrote:
>
> > That works. I'm uncertain whether we should go and skip compiler tests
> too,
> > if the C compiler and OS versions are the same. Probably not, since a
> > different C RTL version will change things.
>
>I would say not to skip compiler checks. In fact one recent bug I found
>was on a newly deployed system that had DECC installed hence configure.com
>was able to determine that CC/DECC was the compiler, but the (admittedly
>redundant) first check in subconfigure.com failed and left me set def'ed
>to the [.UU] subdirectory. The problem was that the LMF license for
>DECC had not been loaded yet, hence one couldn't compile anything in C.
>It is a bug that configure.com didn't spot that problem.
We probably ought to put a check in immediately after we've chosen the
compiler to make sure it works. That'll pick up all manner of
install/config issues.
> > > > question to make it skip the manifest check as well. (That should speed
> > > > things up for multiple rebuilds, especially on VAXen) If we're lucky we
> > >
> > > "-m" : skip the MANIFEST check to see that all files are present
> >
> > I see someone snuck that in for 5.6.0. Keen! I'd remembered the issue
> > coming up, but I missed the fix.
>
>The logic C<$ IF ;$ ENDIF> to skip the check has been in there for ages,
>since folks on this list complained about how slow it was back several
>years ago. It was at one time default off.
Yeah, but the help text only went in as of 5.6.0. (Just checked my 5.005_03
install) Glad it's there.
At some point README.VMS needs rewriting, I think. Things have changed a
bunch since it was first done.
>But if you think it is slow on a VAX try the same MANIFEST check under
>MPW on a 68k Mac - ugghhh! (to top it off MPW regexps are quite weird).
Gack. Sounds worse than MSCP-served RA60s...
Dan
--------------------------------------"it's like this"-------------------
Dan Sugalski even samurai
[EMAIL PROTECTED] have teddy bears and even
teddy bears get drunk
