testing more than one configuration on a single build machine

Gabor Szabo wrote:

Hi,

I have recently signed up to run smoke testing of PostgreSQL.
As I understand each animal has a set of configuration that is represented as
icons on the status page.

I wonder how would be it possible to run several configurations on the
same machine? I mean with or without debug, with or without
thread-safety etc.

This isn't really the place to ask about running buildfarm - it has its
own mailing list to which you should have been subscribed.

A single animal really shouldn't run multiple configurations, because it
makes their history much less valuable. If you want to run multiple
configurations you should register multiple buildfarm members. There is
already support for running several animals on a single host/repo and
this will be improved in a release that I will be preparing RSN.

cheers

andrew

Gabor Szabo wrote:

Hi,

I have recently signed up to run smoke testing of PostgreSQL.
As I understand each animal has a set of configuration that is represented as
icons on the status page.

I wonder how would be it possible to run several configurations on the
same machine? I mean with or without debug, with or without
thread-safety etc.

the usual way to do this is registering different animals for each
configuration.

Stefan

On 8/30/07, Andrew Dunstan <andrew@dunslane.net> wrote:

This isn't really the place to ask about running buildfarm - it has its
own mailing list to which you should have been subscribed.

I am sorry.
Do you mean this?
http://lists.pgfoundry.org/mailman/listinfo/pgbuildfarm-members
I could not find it mentioned in the instructions.

A single animal really shouldn't run multiple configurations, because it
makes their history much less valuable. If you want to run multiple
configurations you should register multiple buildfarm members. There is
already support for running several animals on a single host/repo and
this will be improved in a release that I will be preparing RSN.

Then no thanks. That seems to be too much of hassle to register 2^12 = 4096
animals.It is probably not useful either to test all the permutations
of possible
configurations.

Is there a recommendation which set of configuration might be more important
in finding problems?

Gabor