Re: multi-platform, multi-locale regression tests

From: "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov>
To: <markus(at)bluegap(dot)ch>
Cc: <peter_e(at)gmx(dot)net>,<david(at)kineticode(dot)com>, <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: multi-platform, multi-locale regression tests
Date: 2010-11-12 14:28:27
Message-ID: 4CDCFAAB0200002500037698@gw.wicourts.gov
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Markus Wanner wrote:

> Well, there are three kinds of reporters: StreamReporter,
> TapReporter and CursesReporter. By default, either curser or stream
> is chosen, depending on whether or not dtester thinks its stdout is
> a terminal or not.

> The CursesReporter moves up and down the lines to write results to
> concurrently running tests. It's only useful on a terminal and
> certainly gets confused by anything that moves the cursor (which a
> plain 'print' certainly does).

Ah, well that explains some problems I've had with getting my output
to behave quite like I wanted! Thanks for that summary! I'm pretty
sure I've been getting the CursesReporter; I'll switch to
TapReporter.

> The best solution would probably be to allow the reporters to write
> out comment lines. (However, due to the ability of running tests
> concurrently, these comment lines could only be appended at the
> end, without clear visual connection to a specific test. As long as
> you are only running on test at a time, that certainly doesn't
> matter).

Not sure what the best answer is for Curses -- would it make any
sense to output a disk file with one of the other formats in addition
to the screen, and direct detail to the file? Perhaps a separate
file for each test, to make it easy to keep comments associated with
the test? (Just brainstorming here.)

-Kevin

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Andrew Dunstan 2010-11-12 14:39:35 Re: We need index-only scans
Previous Message Bruce Momjian 2010-11-12 14:17:20 Re: We need index-only scans