On Fri Nov 14 11:42:02 2008, bernhard wrote:
> On Mo. 16. Jun. 2008, 16:50:13, coke wrote:
> > On Wed Jan 16 03:41:56 2008, [EMAIL PROTECTED] wrote:
> > > While the fix to my particular problem is simple enough, it is
> > apparent
> > > that there's enough bit rot in tools/util/smokeserv-server.pl that
> > it
> > > needs a refactoring.  And since there are no unit/regression tests
> > for
> > > this program, it's going to need a good phalanxing.
> 
> > We should probably defer this until post-smolder; we may be able to
> > just drop smoke at that
> > point.
> 
> Is the smoke server still used?

Yes.  A crude search indicates 83 submissions so far this month.

> Can support for the smoke server be dropped?

Yes, but we need to put it through some kind of deprecation cycle before
dropping it entirely.

On the one hand:  I had hoped to refactor the smokeserver code to make
it more user-friendly and to keep it from rejecting perfectly valid
reports.  But coke encouraged me to defer that until after the Smolder
server was established.  We've had Smolder for 6+ months now and, while
it doesn't satisfy my every wish and need, it doesn't reject my reports.
 So I find that I haven't looked at smoke.parrotcode.org for months.

On the other hand:  (a) People are still submitting smoke reports; and
(b) AFAICT the smolder server only reports results of Parrot's 'make
test', where as the smoke server can report tests of languages.  So an
immediate switch to smolder-only would cause a loss in functionality.

Hence the need for a deprecation plan and cycle.  Suggestions?

Thank you very much.
kid51



Reply via email to