Craig A. Berry <[EMAIL PROTECTED]> wrote:
> At 8:34 AM -0500 12/19/00, Charles Lane wrote:
> >Here's a modified patch that incorporates your suggestions. Builds
> >and tests okay with bleadperl @7979 (I haven't tried fighting with the
> >more recent ones yet).
> Just to keep things interesting, I dropped this into 5.6.1-trial1.
> It was failing everything from within the test harness, but all the
> tests I tried individually ran fine. The differences between
> test.com in bleadperl and the trial didn't seem significant until I
> noticed that 5.6.1 doesn't make its own copy of vmspipe.com to do the
> tests, so it depends on the temporary version, but the code that
> creates the temp version on the fly has not been updated.
Damn! I knew that would bite us sometime. It always bothers me when
one has two copies of something that have to be kept in sync manually.
> The patch
> below takes care of that; after this I get "All tests successful."
> After Sarathy gets back from vacation we should badger him to include
> the test.com patch in 5.6.1 as it does really speed things up.
We should get the badgering in the queue now, so that (hopefully) he's
not overbadgered by the time he sees it.
> BTW, sys$output is not supposed to be user-mode?
No, it's not. DCL apparantly handles sys$output a bit differently
than it does sys$error: when you assign to sys$output it does an open,
but only if the assign isn't user-mode. This doesn't seem to happen
for sys$error.
We need this for things like:
$isvax = `\$ Write Sys\$Output F\$GetSyI(\"HW_MODEL\")` <= 1024;
used by gen_shrfls.pl.
I'll apply your patch to vms.c and generate a new patch incorporating all
changes.
--
Drexel University \V --Chuck Lane
======]---------->--------*------------<-------[===========
(215) 895-1545 _/ \ Particle Physics
FAX: (215) 895-5934 /\ /~~~~~~~~~~~ [EMAIL PROTECTED]
