On 03/29/2014 04:47 PM, Tom Lane wrote:
> Noah Misch <noah@leadboat.com> writes:
>> On Sat, Mar 29, 2014 at 09:32:06AM -0400, Tom Lane wrote:
>>> I'm confused. I do not see a single buildfarm failure that appears
>>> attributable to that patch.
>> Here are some:
>> http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=reindeer&dt=2014-03-29%2006%3A00%3A03
> Hm, I was expecting that any failures would be in the main or contrib
> checks. I guess ecpg is a couple directory levels further down though.
>
> In this particular example the path is only about 7 characters too long,
> but I suppose trying to shave off a few characters isn't really going
> to be a robust solution.
>
>
Since you have one of the offending buildfarm machines (dromedary) why
not try this: in the build_env section of the config file, add
PG_REGRESS_SOCK_DIR => "/Users/buildfarm/bf-data/$branch/",
If it works I'll have the buildfarm script default that to a slightly better location in the release I'm about to make.
cheers
andrew
>
>