Re: Regression failure on pika caused by CLUSTER rewrite

Le 15 févr. 2010 à 12:52, Greg Stark <stark@mit.edu> a écrit :

> In looking through the build farm wreckage caused by fsyncing
> directories I noticed this interesting failure on pika:
>
> ================== pgsql.13659/src/test/regress/regression.diffs
> ===================
> *** /home/pgbuildfarm/workdir/HEAD/pgsql.13659/src/test/regress/
> expected/cluster.out    Wed
> Feb  3 00:16:38 2010
> --- /home/pgbuildfarm/workdir/HEAD/pgsql.13659/src/test/regress/
> results/cluster.out    Wed
> Feb  3 19:19:06 2010
> ***************
> *** 453,455 ****
> --- 453,457 ----
>  DROP TABLE clstr_2;
>  DROP TABLE clstr_3;
>  DROP USER clstr_user;
> + ERROR:  role "clstr_user" cannot be dropped because some objects
> depend on it
> + DETAIL:  owner of table pg_temp_9.clstr_temp
>

Hi

I think that was fixed some time ago (see "Fix timing-sensitive
regression test result..." commit by Tom on 02/03). But pika suffered
some connectivity issues then and is only now building current HEAD
again.

Regards,

Rémi Zara