Merge lp:~lifeless/bzr/subunit into lp:bzr

[Robert Collins]

Use subunit when running tests in PQM.

[Robert Collins (lifeless)]

This will turn on subunit in PQM again, for your progress bar and debugging failures edification. I have checked and it exits with non-zero on failure.

[Andrew Bennetts (spiv)]

Seems ok, given that you've tested it exits properly :)

review: Approve

[John A Meinel (jameinel)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Andrew Bennetts wrote:
> Review: Approve
> Seems ok, given that you've tested it exits properly :)

I believe this is currently blocked from landing:
# Generate a stream for PQM to watch.
python2.4 -Werror -O ./bzr selftest --subunit | tee selftest.log
bzr: ERROR: subunit not available. subunit needs to be installed to use
- --subunit.
# Check that there were no errors reported.
subunit-stats < selftest.log
/bin/sh: subunit-stats: not found
make: *** [check-nodocs] Error 127

That is what I saw at the very beginning of a new run, I assume that is
the tail of the previous run. Hey, at least it blocked when it couldn't
succeed. :)

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkuC2zMACgkQJdeBCYSNAANgSACeMFoW8B+Uagai3+ZoS15R0kFH
Dl4AoJTPPLlQIAXN4T5LhuOB/DywgsSO
=KBG0
-----END PGP SIGNATURE-----

[Robert Collins (lifeless)]

On Mon, 2010-02-22 at 19:30 +0000, John A Meinel wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Andrew Bennetts wrote:
> > Review: Approve
> > Seems ok, given that you've tested it exits properly :)
>
> I believe this is currently blocked from landing:
> # Generate a stream for PQM to watch.
> python2.4 -Werror -O ./bzr selftest --subunit | tee selftest.log
> bzr: ERROR: subunit not available. subunit needs to be installed to use
> - --subunit.
> # Check that there were no errors reported.
> subunit-stats < selftest.log
> /bin/sh: subunit-stats: not found
> make: *** [check-nodocs] Error 127
>
> That is what I saw at the very beginning of a new run, I assume that is
> the tail of the previous run. Hey, at least it blocked when it couldn't
> succeed. :)

Yep, thats the size of it. Following up wit losas.

-Rob

[Martin Pool (mbp)]

back to wip pending build environment dependencies being sorted out

[Martin Pool (mbp)]

that's rt 37636

=== modified file 'Makefile'
--- Makefile	2010-02-18 02:11:03 +0000
+++ Makefile	2010-02-20 09:39:14 +0000
@@ -39,7 +39,10 @@
 check: docs check-nodocs
 
 check-nodocs: extensions
-	$(PYTHON) -Werror -O ./bzr selftest -1 $(tests)
+	# Generate a stream for PQM to watch.
+	$(PYTHON) -Werror -O ./bzr selftest --subunit $(tests) | tee selftest.log
+	# Check that there were no errors reported.
+	subunit-stats < selftest.log
 
 # Run Python style checker (apt-get install pyflakes)
 #

=== modified file 'NEWS'
--- NEWS	2010-02-18 04:59:30 +0000
+++ NEWS	2010-02-20 09:39:14 +0000
@@ -105,6 +105,9 @@
   dependency regressions.
   (Martin Pool)
 
+* PQM will now run with subunit output. To analyze a PQM error use
+  tribunal, or cat log | subunit-filter | subunit2pyunit. (Robert Collins)
+
 * Stop sending apport crash files to ``.cache`` in the directory from
   which ``bzr selftest`` was run.  (Martin Pool, #422350)