develooper Front page | perl.perl5.porters | Postings from September 2003

Re: 5.8.2 and onwards

Thread Previous | Thread Next
From:
Nicholas Clark
Date:
September 27, 2003 06:14
Subject:
Re: 5.8.2 and onwards
Message ID:
20030927141447.O30988@plum.flirble.org
On Sat, Sep 27, 2003 at 01:24:00PM +0200, Abe Timmerman wrote:
> Op een zonnige herfstdag (Saturday 27 September 2003 12:28), schreef Nicholas 
> Clark:

> > Maybe controversial and undiplomatic, but
> >
> > 1: reduce traffic on p5p by not sending 100% pass smoke reports
> 
> Would you like me to think of a way to try and manage that from Test::Smoke?

Yes please, if you could

I can think of two options

The simpler would be to provide a configure option to Cc: an address only
if the smoke has failures in the matrix (for some value of fail)
I think that in itself this would be useful (but only experience will tell)

The more sophisticated is if the smoke can store the state of the previous
results matrix, and only report if the state changes (probably including
changes in which configurations failed, maybe not down to the level of
the individual tests. Not sure).

This would mean that if a platform breaks, and then 8 smokes later we
fix it, we'd only get 2 messages here (although full results would still
be on the smokers list and Alian's smoke database)
The danger is that any system that yo-yos in and out of failure will now
generate 2 messages for each round trip, rather than 1 (for the first
scheme)

There may be other ways to do it.

Nicholas Clark


Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About