[RFE] option to "make check" be more verbose When I run the tests, and they pass I get something like this: sh-3.1$ pwd /cygdrive/c/builds/trunk/mozilla sh-3.1$ cd ff-debug/browser/components/places/tests/ sh-3.1$ make check ../../../../dist/bin/test_all.sh ../../../../dist/bin/test_places ../../../../dist/bin/test_places/test_bookmarks.js: PASS ../../../../dist/bin/test_places/test_livemarks.js: PASS Is there a way we could add a "make checkverbose" target so that we'd see more of what's going on? I was thinking about http://lxr.mozilla.org/seamonkey/source/tools/test-harness/xpcshell-simple/head.js and doing a dump() on all do_check_*() functions. robert, what do you think?

I think that's fine if-wanted, but the tests themselves will have to be rewritten to handle the extra verbosity. An integer value for verbosity a la xUnit might be nice. e.g., 0 = terse, 1 = standard 2+ = verbose ?

(In reply to comment #1) > I think that's fine if-wanted, but the tests themselves will have to be > rewritten to handle the extra verbosity. An integer value for verbosity a la > xUnit might be nice. e.g., 0 = terse, 1 = standard 2+ = verbose ? > Let's use the same thing mochikit does-- a Log4J approach. DEBUG, INFO, ERROR, etc. Test failures are logged at ERROR level.

make check output on the mac is different from the other platforms. If I run make -k check in my objdir, I get notice that deps are being updated, but it doesn't tell me where. Linux and Windows both report which directories are being entered and left.

(In reply to comment #0) > http://lxr.mozilla.org/seamonkey/source/tools/test-harness/xpcshell-simple/head.js > doing a dump() on all do_check_*() functions. Just had this idea too while modifying a test ;-) (In reply to comment #1) > I think that's fine if-wanted, but the tests themselves will have to be > rewritten to handle the extra verbosity. Let's morph this bug: add dump()s to what we have now (only), which will go to the individual log files, then leave the rest to bug 468196. (In reply to comment #3) > make check output on the mac is different from the other platforms. If I run > make -k check in my objdir, I get notice that deps are being updated, but it > doesn't tell me where. Linux and Windows both report which directories are > being entered and left. That's unrelated to this bug, is it?

The output is still a little raw, but this is a major enhancement yet. [Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9.2a1pre) Gecko/20090531 Minefield/3.6a1pre] (mozilla-central-win32-unittest/1243808976) (W2Ksp4) (http://hg.mozilla.org/mozilla-central/rev/ab395a1916be) 628 TEST-PASS = 579 with checks + 49 without checks.

Comment on attachment 381133 [details] [diff] [review] (Av1) Log check results [Checkin: See comment 8 & 9] I think this looked pretty good on a quick pass. coop, do you have time to give this a once-over?

Comment on attachment 381133 [details] [diff] [review] (Av1) Log check results [Checkin: See comment 8 & 9] > function do_test_pending() { >- dump("*** test pending\n"); >- _tests_pending++; >+ dump("TEST-INFO | (xpcshell/head.js) | test " + (++_tests_pending) + >+ " pending\n"); > } My only concern is incrementing _tests_pending while we're concatenating. I'd prefer that as two distinct statements.

Comment on attachment 381133 [details] [diff] [review] (Av1) Log check results [Checkin: See comment 8 & 9] http://hg.mozilla.org/mozilla-central/rev/07971f68e728 (Av1a) Log check results

Comment on attachment 381133 [details] [diff] [review] (Av1) Log check results [Checkin: See comment 8 & 9] http://hg.mozilla.org/releases/mozilla-1.9.1/rev/3bab31a7104f after fixing context for { patching file testing/xpcshell/head.js Hunk #2 FAILED at 40 1 out of 5 hunks FAILED patching file testing/xpcshell/runxpcshelltests.py Hunk #1 FAILED at 14 Hunk #2 FAILED at 208 2 out of 2 hunks FAILED }

verified1.9.1, per bug 496776 comment 4.