Re: 2018-03 CFM - Mailing list pgsql-hackers

On Sat, Mar 3, 2018 at 1:18 AM, Aleksander Alekseev
<a.alekseev@postgrespro.ru> wrote:
>> You do realize we have the actual source database available, I hope? Since
>> it's our own system... There is no need to scrape the data back out -- if
>> we can just define what kind of reports we want, we can trivially run it on
>> the source database. Or if we want it more often, we can easily make a
>> webview for it. It's basically just a "map this URL to a SQL query"...
>
> I don't think commitfest.cputube.org has the SQL data on whether patch
> pass the tests. It just displays SVG images from travis-ci.org. Also
> unfortunately both commitfest.postgresql.org and commitfest.cputube.org
> currently don't have any kind of public API and don't allow to export
> data, e.g. in CSV or JSON.
>
> I guess it would be nice if both services supported export, in any
> format, so anyone could build any kind of reports or automation tools
> without parsing HTML with regular expressions or depending on other
> people.
>
> If I'm not mistaken, there was a discussion regarding public APIs.
> I wonder what prevents adding it, at least a simple export of everything.
> After all, it is indeed just mapping URL to a SQL query. For instance,
> this one:
>
> select array_to_json(array_agg(row_to_json(tbl))) from tbl;

Hi Aleksander,

My current plan is to propose that we post automated apply/build
results into the Commitfest app itself so that it can show them, and
then shut down commitfest.cputube.org.  I don't have all the details
figured out yet and I won't be working on a proposal until after this
Commitfest is over.  If you have ideas on how it should receive, store
and show them, please clone the app (pgcommitfest2.git?) and start a
thread (maybe on pgsql-www?).  Here's a question for you to ponder:  I
think there should probably be one "apply" result (success/fail, and a
URL to find the log), but should there be one or many "build/test"
results?  It could show results from different OSes separately (Travis
macOS, Travis Ubuntu, AppVeyor Windows, ...), and it could track more
than one recent result (useful for seeing things that used to pass and
now fail or vice versa).

The reason commitfest.cputube.org is so limited, not integrated yet
and running on a strange domain name is because I decided to build an
independent proof-of-concept first.  I imagined that negotiating with
many busy people on how such a thing should work and what would even
be possible first would be more likely to stall, based on previous
experiences with humans :-D

-- 
Thomas Munro
http://www.enterprisedb.com