Because the new-style tests are intrustive,
the exception that CppUnit throws on assertion
failures is caught and processed with the
application logic, which is far from ideal,
because it's very difficult to find the
cause of failure.
What we'd like is a way to control what happens
when an test assertion fails, such that we can
properly log/print the failure, and even break
in the debugger.
The new macros allow us to control the behavior
at compile-time and have added flexibility.
For now, they log an assertion failure before
invoking the CPPUNIT macro, and support a
compile-time directive to assert, which is
useful for breaking in the debugger.
Change-Id: If464ba246e3ec747f31496a4215cb73ef735dfaf
Reviewed-on: https://gerrit.libreoffice.org/c/online/+/87625
Tested-by: Jenkins CollaboraOffice <jenkinscollaboraoffice@gmail.com>
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
This restores commit 5c8a224e2d, which
disabled the unstable testReloadWhileDisconnecting testcase. Later this
was reverted by commit 4fc7bc76db, but
nothing was really done to improve the stability, so go back to disabled
for now. (Till somebody has time to debug & fix the root cause here.)
Change-Id: I26ea40a81cf9e2673f2397221c85e6f9dedc4052
Reviewed-on: https://gerrit.libreoffice.org/c/online/+/88669
Tested-by: Jenkins CollaboraOffice <jenkinscollaboraoffice@gmail.com>
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
So we can see how stable the cypress tests are. Now other unit
tests under tests folder are failing randomly, so it's hard to
monitor the cypress tests.
Enable one unstable unit test, which was disabled for the same
reason.
Change-Id: Ib1646de1647c9a2bdb769bbb6b92a5794dd7a598
Reviewed-on: https://gerrit.libreoffice.org/c/online/+/88559
Tested-by: Tamás Zolnai <tamas.zolnai@collabora.com>
Reviewed-by: Tamás Zolnai <tamas.zolnai@collabora.com>
So that they are in-process, which means it's easier to debug when they
fail.
Also, in UnitLoad, give up on trying to avoid the sleep after
disconnecting, it seems the old condition was not reliable.
Change-Id: I972a3319887a70eeea2585104ed1e762830ca505
This was originally moved in commit
b97789eb76 (Convert paste testcase to a
new-style one, 2019-10-18), then
194db8ed45 (test: improve stability of a
number of tests, 2019-10-12) restored a duplicate of this, probably due
to a mismerge.
Change-Id: Ie76266ddb85c58bb2370640025b91af980656908
There are numerous race conditions between issuing
action and getting the events. This causes a
mismatch, such that the events from SelectAll
is received when waiting for the events for a
subsequent action (say, Delete).
To make matters worse, sometimes Core issues
an invalidation whilst sending the events for
an action, and this completely messes our
accounting of what events we expect.
This latter could also be an issue with real
clients, however it's less likely to be
disruptive. Still, it is possible that
the client doesn't get key events
because Core had a hiccup, which breaks
the tests.
For the tests at least, the solution is to
re-issue the last action, such that we make
sure that we move forward only when the correct
event is received, or we timeout waiting.
The end results is that tests now don't fail
nearly as often as they used to.
Reviewed-on: https://gerrit.libreoffice.org/80894
Reviewed-by: Andras Timar <andras.timar@collabora.com>
Tested-by: Andras Timar <andras.timar@collabora.com>
(cherry picked from commit 050403daf0)
Change-Id: Ib1a658846061990649987ca083915a49c1fb092a
Reviewed-on: https://gerrit.libreoffice.org/81565
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
This was originally added in commit
c2c7982e5e (Add TIFF load testcase,
2019-07-04), and still fails without the core.git fix.
Change-Id: I798b82ea8212dfeafc162e4b73f6fff707ac3de2
After commit 8e118889cb (wsd: leaflet:
support statusupdate: messages to sync clients, 2019-08-17), which added
a new hiddenparts= key, but forgot to adapt tests accordingly.
Change-Id: Ia5baefff37b88f5b173d65ae80a126f94b420cb0
It seems to me that in this case the test was poor, if other
invalidation types are expected, then this one should be fine, too.
Change-Id: Iae7597eef081acdfb5d6f9152747204a71bd2b66
Fails with core.git commit ad8f4c4432e0c5c052b6f87977d1ad8663debe09 (vcl
lok: pre-load graphic libraries, 2019-07-04) reverted, which fixed TIFF
loading in the preload case (running in a chroot, no code available in
the chroot).
Change-Id: Ic46da0abcb13c4dd6fc1ce697b93e0b3062077d7
This test only passed when core.git was built in a way so that '#if
OSL_DEBUG_LEVEL > 0' was true (debug, dbgutil builds). Re-generate this
against a product build, since 2 other tests already have references
like that.
We already have infrastructure for stripping <desc> XML elements from
the actual result.
Change-Id: I9cdd53bf9625266d94313a90946270bfc266bd47
... debug core.git.
SVGActionWriter::ImplWriteActions() in core.git writes additional <desc>
XML elements, guarded with a '#if OSL_DEBUG_LEVEL > 0' block.
Filter these out, so the reference SVG will match both product and debug
builds.
Change-Id: Iba3fb25af206c70d5a4ff5801a934dcfd74704de
DynamicStruct is also more recent and is missing
from older distros and versions of Poco, so avoid
using it just for that.
Change-Id: I465fbbc4350101d382877bd42a1704e4e7827102
Unfortuantely it still fails, but this is not regression. Now
with the new patch in Core the exception is caught and so at least
the binary survives (and the API returns 0).
New unit-test added in Core to help track the issue down and fix.
Also, free the memory allocated by the API.
Change-Id: I5d788a2ee0383de1c323af4cd6b39b8615a35baf
