153259cb10
tests: also get rid of smartmatch in complete-run.pl
2013-09-24 06:37:40 +02:00
3ca8299637
complete-run: clarify the message about missing ../i3 (Thanks Marcos)
2012-12-20 23:50:50 +01:00
773654dbb8
complete-run: run 000-load-deps as early as possible
2012-11-13 21:04:13 +01:00
f9c903ba62
complete-run.pl: Check for missing executables
2012-09-20 11:37:49 +02:00
be6190a516
complete-run: check whether Xdummy dies, add a flag to keep the Xdummy output
2012-08-18 16:27:00 +02:00
b01d45e027
complete-run: handle bailouts
2012-08-12 13:46:54 +02:00
d92626bc62
complete-run: enable autoflush (useful for debugging)
...
I used the following command to find hanging workers:
sort latest/complete-run.log|cut -f 2 -d ' '|uniq -c|grep -v '^[ \t]*2'
2012-08-12 00:20:52 +02:00
077e021e26
tests: implement --xtrace in complete-run.pl
2012-08-04 03:12:06 +02:00
0f10ccdf12
Implement fake-outputs option (cmdline, cfg) for multi-monitor testing
...
This kills the dependency on xdmx and makes the testsuite simpler
and more flexible (in the output sizes / configurations).
2012-04-09 14:28:36 +02:00
d22458cde0
complete-run: make --valgrind, --strace, --coverage-testing work again
2012-01-27 21:36:40 +00:00
dd9743b272
tests: Implement multi-monitor tests using Xdmx
2011-12-26 20:48:57 +01:00
69b143e5ca
complete-run: automatically display test output when running a single test
2011-12-17 11:20:32 +00:00
95508c3469
complete-run: fix uninitialized warning
2011-12-17 11:20:16 +00:00
81f4b6fc1a
complete-run: don’t parallelize more than the number of tests
2011-12-17 11:19:31 +00:00
1c72e8b69e
Make complete-run store the timings, schedule tests and print time estimate
...
Yay for self-optimizing software and time estimates
2011-12-16 22:00:41 +00:00
9a7d7919a6
testcases: introduce TestWorker.pm
...
instead of executing a new perl interpreter (via TAP::Parser)
each time we start a testfile, fork a TestWorker for each display.
Each worker preloads i3test via 'require', blocking waits on its ipc
to get a new filename, forks itself upon arrival and 'do'es this
testscript.
2011-12-04 14:14:20 +01:00
55c474d864
testcases: move Xdummy cleanup code to StartXDummy
2011-12-04 14:14:20 +01:00
43b8b8356d
complete-run: add cleanup sighandler
2011-12-04 14:14:20 +01:00
2d188bfc9b
complete-run: reorder code to make code flow more clear
2011-12-04 14:14:20 +01:00
6a5ca32c5a
complete-run: die if X11::XCB couldn’t connect to a display
2011-12-04 14:14:20 +01:00
8349190e09
testcases: always start i3 through i3test::launch_with_config
...
this will pave the way to just call BAIL_OUT and stop the currently
running testfile if i3 died for whatever reason.
2011-12-04 14:14:20 +01:00
423b891995
StartXdummy.pm: make Xdummy startup a bit more robust
2011-12-04 14:14:20 +01:00
af793c9030
complete-run: wait for all jobs to be finished
2011-11-21 19:20:52 +00:00
3abd7ab073
complete-run: close all fds except for 0, 1, 2
...
running in a VIM subshell leads to one more fd, for example
2011-11-20 10:51:21 +00:00
77a9e4b18f
complete-run: implement --strace, make --valgrind log to test-specific file
2011-11-15 23:30:17 +00:00
2c48280017
complete-run: remove Carp::Always, slipped in when debugging
2011-11-09 22:41:56 +00:00
316318c1fc
complete-run: don’t print long logfile paths, they are useless by now anyway
2011-11-09 22:38:29 +00:00
7633aa67a1
complete-run: log verbose output, display statuslines instead
2011-11-09 23:00:10 +01:00
afc488021f
complete-run.pl: automatically start Xdummy instances unless -d is specified
...
This makes running the testsuite incredibly easy:
$ ./complete-run.pl
:)
2011-11-07 23:04:45 +00:00
8b887e8447
complete-run: Bugfix: return condvar when $dont_start is true
2011-11-07 22:38:06 +00:00
0615cb3595
complete-run.pl: implement --help
2011-11-07 21:21:51 +00:00
fdf7b1706c
complete-run: Bugfix: Don’t call recv inside a callback when cleanly exiting i3
2011-11-07 20:56:36 +00:00
dbd6440432
complete-run: Use Carp::Always to get nice stacktraces in case of errors
2011-11-07 20:56:04 +00:00
1c0d69d4e6
complete-run: implement --valgrind
2011-11-07 20:53:49 +00:00
389fdcf79f
complete-run: pass outdir (not only logpath) to activate_i3()
2011-11-07 20:52:14 +00:00
1056ecc885
complete-run: eliminate dependency on EV
2011-10-05 23:52:19 +01:00
27a38a3917
complete-run: explicitly state why we need to overwrite SIGCHLD
2011-10-05 23:29:58 +01:00
b9224634dd
tests: eliminate Try::Tiny
2011-10-05 23:21:23 +01:00
689f3b8cf7
tests: Eliminate IO::Scalar
2011-10-05 23:17:09 +01:00
bd33c09845
tests: move i3test.pm from t/lib to lib/
2011-10-04 23:39:36 +01:00
8d9b4c3c04
tests: launch_with_config: use socket activation
2011-10-04 23:33:38 +01:00
6c7c4d52d0
tests: Refactor the socket activation into lib/SocketActivation.pm
2011-10-04 23:30:30 +01:00
a94ec5ee4e
tests: re-order dependencies in complete-run, make clear which are shipped with Perl
2011-10-04 20:10:11 +01:00
53121c746c
tests: remove unused dependency Proc::Background from complete-run
2011-10-04 20:05:12 +01:00
aa65b721b8
tests: eliminate dependency on the DateTime module by using POSIX::strftime
2011-10-04 20:03:08 +01:00
20d6ae4110
tests: make complete-run use POSIX::close(3) instead of reserving a file descriptor when starting
...
This makes it possible to run complete-run.pl with a "tainted" environment in
which fd 3 (and possibly others) are already present.
2011-10-04 19:58:17 +01:00
1eb011aae1
tests: make sure to leave no tempfiles behind
2011-09-24 16:28:21 +01:00
c3eb9f6c45
tests: complete-run: remove debugging messages
2011-09-24 13:13:26 +01:00
1a438f12ed
tests: complete-run: display time i3 took for starting up
2011-09-24 13:13:09 +01:00
e244a75801
tests: complete_run: directly use X11::XCB instead of ::Connection
...
This saves about 0.5s wallclock time due to not starting up Moose/Mouse.
This is worthwhile when you develop a new feature and you are often invoking
complete_run for one specific test.
2011-09-24 13:09:20 +01:00
Michael Stapelberg