Op een grimmige herfstdag (Sunday 14 November 2004 01:28),schreef Craig A.
Berry:
> At 8:11 PM +0100 11/13/04, Abe Timmerman wrote:
> >I realize i have been sending my smokereports to this list without even
> > asking what the policy/opinion is.
> >
> >So I'll ask now.
> >
> >>From Test::Smoke there are three options:
> >
> >1) always (PASS or FAIL) (current setting)
> >2) only on FAIL
> >3) never
> >
> >Please let me know what you want!
>
> Personally I would vote for always, at least until we get more of the
> kinks worked out of Test::Smoke. For the benefit of vmsperlers who
> may not know what the fuss is about, these automated test reports
> allow the committers to know within a matter of hours how code
> changes checked into the repository affect the test results on a
> whole range of platforms with various configurations. Until now,
> this has not been available on VMS, so we've depended on my manual
> tests to know how things stand.
Thank you for that clear explanation.
Meanwhile I've set up both the VAX and Alpha box to do continuous smoking and
(finaly) automated mailsending. I'll add vmsperl to the recipiants list until
you have enough of it.
> Abe has gotten things pretty far and even seems to be brave enough to
> do a little DCL :-).
:-)
> I think there is some additional work to do,
> though. For example, I don't think Test::Smoke knows how to add
> qualifiers to MMS or MMK, such as with the /MACRO="__DEBUG__"=1 case
> to do a vmsdebug build.
I'll make a new snapshot this weekend. My version already knows about
-"Dusevmsdebug" and /macro=("__DEBUG__=1"), but I must admit that if there
are other qualifiers needed by certain configure options they will not be
set. (I'll need to do more digging in configure.com)
> I also see tests failing when I know they
> pass outside of Test::Smoke. So we are getting close but there is
> more to do.
Yes, this is a problem that exists also on other platforms :-(
We need to root these out but it has proven difficult.
Good luck,
Abe
--
(*) Before the testing posse is on my case :-) let me preempt them by
saying that fake networks don't really help here. You'll be testing
a fake network. That proves that the module worked on a fake network.
That's great. Like drinking decaffeinated.
-- Jarkko Hietaniemi on p5p @ 2001-11-22
