Boost logo

Boost Testing :

From: David Abrahams (dave_at_[hidden])
Date: 2006-09-21 11:20:55


Markus Schöpflin <markus.schoepflin_at_[hidden]> writes:

> David Abrahams wrote:
>
>> The ones from Sep 15 are woefully outdated; they're still showing
>> errors that I've long since fixed and would like to see dropped from
>> the regression report. Surely, by that logic, Bronek[2-6]-v2.zip and
>> COMSOFT.zip must be intolerably old... and "RSI Droid2K V2.zip" is
>> from July 28!!
>
> My regression tests (COMSOFT) are running fine, but it fails to upload the
> results to the FTP server since the noted date. I have no idea what is
> causing this, because it works when I'm doing it manually.
>
>> Is anyone using these results? Have the respective testers stopped
>> testing? Are their platforms/compilers going to be supported for the
>> release?
>
> I think that you're a bit impatient in saying that these results are
> woefully outdated. It's not even a week since the last update.

It's woeful when I think I fixed the bug a week ago but can't get
confirmation.

> It takes more than two day for a full clean regression run to
> complete. And yes, I hope that Tru64 will still be a supported
> platform after the next release.
>
> And I think it's a bit unfair to start complaining _now_, because these
> regression tests (including HP TestDrive) have been running continuously
> since the RC branch was created, mostly without any problems.

I don't see what fairness has to do with it. I'm complaining _now_
because I started noticing that my fixes weren't registering and I've
recently been turning my work life inside out to clear the reports.

Maybe we need a way to tell the system "stop
reporting any failures from _existing_ regressions runs on this
lib/test/whatever," so we only get reports from new regression runs.
We'd need to make sure all those settings were cleared before release,
of course.

> And it would be very kind of you to not simply remove the regression tests
> I'm running without asking first.

Sure thing; I was actually trying to move them to a different
directory so they could be put back if anyone needed them. Didn't
work, though.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com

Boost-testing list run by mbergal at meta-comm.com