platform/upstream/gstreamer.git
8 years agovalidate: Blacklist scrub_forward_seeking.op2b-mpeg2-wave_hd_mxf
Edward Hervey [Fri, 12 Aug 2016 10:30:41 +0000 (12:30 +0200)]
validate: Blacklist scrub_forward_seeking.op2b-mpeg2-wave_hd_mxf

See https://bugzilla.gnome.org/show_bug.cgi?id=764025

8 years agovalidate: use new API when switching track with playbin3
Guillaume Desmottes [Thu, 28 Jul 2016 07:47:42 +0000 (09:47 +0200)]
validate: use new API when switching track with playbin3

Move all the implementations of 'switch-track' to
gst-validate-scenario.c while doing so.

Differential Revision: https://phabricator.freedesktop.org/D1227

8 years agovalidate: reporter: break cyclic references with reports
Guillaume Desmottes [Tue, 31 May 2016 10:32:16 +0000 (12:32 +0200)]
validate: reporter: break cyclic references with reports

My patch fixing monitor leak (15e7f1bbfd84ce2cc5e6420fee2255c2be95e0f6)
introduced a ref cycle between GstValidateReporter and
GstValidateReport.

The reports uses its reporter so it needs a ref on it
to ensure it's stay alive. But reports are owned by
GstValidateReporter and/or GstValidateRunner.

Fix this by not taking a reference on the reporter but instead caching
its name.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D1029

8 years agovalidate: turn GstValidateReport to a mini object
Guillaume Desmottes [Thu, 26 May 2016 12:02:45 +0000 (14:02 +0200)]
validate: turn GstValidateReport to a mini object

It handles refcounting for us and will enable automatic leak checks when
using the 'leaks' tracer.

Differential Revision: https://phabricator.freedesktop.org/D1233

8 years agovalidate: inherit from GstObject instead of GObject
Guillaume Desmottes [Thu, 26 May 2016 10:32:16 +0000 (12:32 +0200)]
validate: inherit from GstObject instead of GObject

This allow us to use to 'leaks' detector to check if those objects are
leaked.

Differential Revision: https://phabricator.freedesktop.org/D1232

8 years agovalidate: call gst_deinit() after gst_validate_deinit()
Guillaume Desmottes [Fri, 20 May 2016 12:46:19 +0000 (15:46 +0300)]
validate: call gst_deinit() after gst_validate_deinit()

This allows validate to clean up before the 'leak' tracer list leaked
objects.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D1231

8 years agovalidate: use MAY_BE_LEAKED flag
Guillaume Desmottes [Fri, 20 May 2016 12:44:20 +0000 (15:44 +0300)]
validate: use MAY_BE_LEAKED flag

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D1230

8 years agovalidate: fix pad leaks
Guillaume Desmottes [Mon, 30 May 2016 13:42:24 +0000 (15:42 +0200)]
validate: fix pad leaks

Pads returned using the playbin get-{audio,video}-pad are reffed.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D1027

8 years agovalidate: reporter: break cyclic references with reports
Guillaume Desmottes [Fri, 27 May 2016 13:37:00 +0000 (15:37 +0200)]
validate: reporter: break cyclic references with reports

My patch fixing monitor leak (15e7f1bbfd84ce2cc5e6420fee2255c2be95e0f6)
introduced a ref cycle between GstValidateReporter and
GstValidateReport.

The reports uses its reporter so it needs a ref on it
to ensure it's stay alive. But reports are owned by GstValidateReporter and/or
GstValidateRunner.
The best way I found to break this cycle is to introduce this purge
method. It's not great but the design is a bit tricky.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D1029

8 years agovalidate: reporter: prevent usage of destroyed runner
Guillaume Desmottes [Fri, 27 May 2016 12:36:44 +0000 (14:36 +0200)]
validate: reporter: prevent usage of destroyed runner

Fix crashes.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D1028

8 years agovalidate: runner: prevent hash table modifications while iterating
Guillaume Desmottes [Fri, 27 May 2016 11:23:48 +0000 (13:23 +0200)]
validate: runner: prevent hash table modifications while iterating

A GHashTableIter is invalided if the hash table is modified while we are
iterating. Prevent this by taking the runner lock.

Fix assertion warnings with
validate.file.transcode.to_vorbis_and_vp8_in_webm.Sintel_2010_720p_mkv_srt

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D1026

8 years agovalidate: Add jpeg as known format
Thibault Saunier [Fri, 29 Jul 2016 19:52:48 +0000 (15:52 -0400)]
validate: Add jpeg as known format

8 years agovalidate: Fix testsuite after additional check for buffer DISCONT flag
Thibault Saunier [Fri, 29 Jul 2016 17:27:23 +0000 (13:27 -0400)]
validate: Fix testsuite after additional check for buffer DISCONT flag

8 years agodata: Fix make distcheck.
Mathieu Duponchelle [Tue, 19 May 2015 13:53:06 +0000 (13:53 +0000)]
data: Fix make distcheck.

by distributing newly-added files.

Reviewed-by: Guillaume Desmottes <guillaume.desmottes@collabora.co.uk>
Differential Revision: https://phabricator.freedesktop.org/D185

8 years agovalidate: scenario: deploy setup_sink_props_max_lateness config scenario for valgrind
Wonchul Lee [Thu, 15 Oct 2015 07:29:27 +0000 (07:29 +0000)]
validate: scenario: deploy setup_sink_props_max_lateness config scenario for valgrind

Add to deploy setup_sink_props_max_lateness scenario.
When running gst-validate with valgrind option on the installed package, it fails to find that scenario.

Reviewed-by: Guillaume Desmottes <guillaume.desmottes@collabora.co.uk>
Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D379

8 years agodocs: Fix typo
Wonchul Lee [Mon, 18 Jan 2016 03:53:20 +0000 (03:53 +0000)]
docs: Fix typo

Reviewed-by: Alex Băluț <alexandru.balut@gmail.com>
Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D681

8 years agovalidate:launcher: Add h265, opus and vp9 as known formats
Thibault Saunier [Fri, 15 Jul 2016 12:56:02 +0000 (08:56 -0400)]
validate:launcher: Add h265, opus and vp9 as known formats

8 years agoBack to development
Sebastian Dröge [Wed, 6 Jul 2016 10:51:27 +0000 (13:51 +0300)]
Back to development

8 years agoRelease 1.9.1
Sebastian Dröge [Wed, 6 Jul 2016 10:48:18 +0000 (13:48 +0300)]
Release 1.9.1

8 years agovalidate: Fix usage for non-standard http server port
Edward Hervey [Mon, 4 Jul 2016 14:16:25 +0000 (16:16 +0200)]
validate: Fix usage for non-standard http server port

Translate the various stream_info URI from the standard port to the
specified port
Make the port option an integer option

8 years agomedia-descriptor: Fine-tune stream-id checking
Edward Hervey [Mon, 4 Jul 2016 14:12:27 +0000 (16:12 +0200)]
media-descriptor: Fine-tune stream-id checking

Only expect fully identical stream-id from URI which are not local files
nor from our local http server.

Fixes issues with non-default http server port

8 years agovalidate: Use presence of STREAM_START to detect track switching
Edward Hervey [Mon, 23 Nov 2015 14:11:58 +0000 (15:11 +0100)]
validate: Use presence of STREAM_START to detect track switching

When doing a track switch, the only reliable way to detect that it
happened is whether a new STREAM_START arrives.
Relying on a DISCONT buffer is not satisfactory, since there might
not have been an element setting that flag upstream.

Checking whether the first buffer after a STREAM_START has the
DISCONT flag properly set should be done in parallel

8 years agotools: Make the stream-switching action more generic
Edward Hervey [Thu, 5 Nov 2015 09:29:33 +0000 (10:29 +0100)]
tools: Make the stream-switching action more generic

This allows it to handle both playbin and playbin3

8 years agotools: Fix playbin detection
Edward Hervey [Thu, 5 Nov 2015 09:19:22 +0000 (10:19 +0100)]
tools: Fix playbin detection

We might be using playbin3 instead

8 years agovalidate: Use decodebin3/uridecodebin3 when specified
Edward Hervey [Mon, 2 Nov 2015 13:42:53 +0000 (14:42 +0100)]
validate: Use decodebin3/uridecodebin3 when specified

Set the USE_PLAYBIN3 environment variable to use those elements instead
of legacy ones.

8 years agotools: New tool to view and compare xunit results
Edward Hervey [Thu, 5 Nov 2015 15:37:17 +0000 (16:37 +0100)]
tools: New tool to view and compare xunit results

8 years agovalidate: fix one more export
Tim-Philipp Müller [Fri, 24 Jun 2016 13:03:48 +0000 (14:03 +0100)]
validate: fix one more export

gst-validate-scenario.c:183:7: error: '_gst_validate_action_type' redeclared without dllimport attribute: previous dllimport ignore

This is also declared in gst-validate-internal.h

8 years agovalidate: fix bogus exports for internal symbols
Tim-Philipp Müller [Fri, 24 Jun 2016 10:48:50 +0000 (11:48 +0100)]
validate: fix bogus exports for internal symbols

Hopefully fixes win32 build.

gst-validate-scenario.c:183:7: error: '_gst_validate_action_type' redeclared without dllimport attribute: previous dllimport ignored
gst-validate-scenario.c:286:1: error: '_action_check_and_set_printed' redeclared without dllimport attribute: previous dllimport ignored
gst-validate-scenario.c:298:1: error: 'gst_validate_action_is_subaction' redeclared without dllimport attribute: previous dllimport ignored
gst-validate-scenario.c:305:7: error: '_gst_validate_action_type_type' redeclared without dllimport attribute: previous dllimport ignored

8 years agovalidate: Fix Validate plugin paths mixup
Thibault Saunier [Mon, 13 Jun 2016 21:00:39 +0000 (17:00 -0400)]
validate: Fix Validate plugin paths mixup

8 years agovalidate: fix typo in debug message
Reynaldo H. Verdejo Pinochet [Fri, 10 Jun 2016 19:55:47 +0000 (12:55 -0700)]
validate: fix typo in debug message

8 years agovalidate: fix some recurring typos
Reynaldo H. Verdejo Pinochet [Thu, 9 Jun 2016 21:30:53 +0000 (14:30 -0700)]
validate: fix some recurring typos

8 years agovalidate: fix validate-transcoding option descriptions
Reynaldo H. Verdejo Pinochet [Thu, 9 Jun 2016 21:22:31 +0000 (14:22 -0700)]
validate: fix validate-transcoding option descriptions

8 years agovalidate: improve run-time option descriptions
Reynaldo H. Verdejo Pinochet [Thu, 9 Jun 2016 20:57:33 +0000 (13:57 -0700)]
validate: improve run-time option descriptions

8 years agovalidate: No execute ON_ADDITION if a previous action has a playback-time
Thibault Saunier [Thu, 9 Jun 2016 20:41:57 +0000 (16:41 -0400)]
validate: No execute ON_ADDITION if a previous action has a playback-time

As this is what user will expect in this case.

For example with this scenario:

    set-state, state=null; playback-time=5
    set-property, target-element-name=dvbsrc0, property-name=delsys, property-value=11
    play;

8 years agovalidate:launcher: Add support for relative path while providing file path
Vineeth TM [Tue, 11 Aug 2015 07:41:20 +0000 (16:41 +0900)]
validate:launcher: Add support for relative path while providing file path

Instead of providing full absolute path while validating the file, should be
able to provide the relative path with respect to the present directory.

https://bugzilla.gnome.org/show_bug.cgi?id=753494

8 years agovalidate: fix monitor leak when doing frame analysis
Guillaume Desmottes [Tue, 24 May 2016 14:05:37 +0000 (14:05 +0000)]
validate: fix monitor leak when doing frame analysis

The monitor returned by gst_validate_monitor_factory_create() was never
unreffed.
Report instances now have to keep a ref, as suggested by the TODO, as
the reporter is no longer leaked.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D1012

8 years agovalidate: redefine default cat for the runner
Guillaume Desmottes [Tue, 24 May 2016 14:05:30 +0000 (14:05 +0000)]
validate: redefine default cat for the runner

The gstvalidate_debug may not be initialized like with the
validate/reporting which was crashing when run with GST_DEBUG=5.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D1004

8 years agovalidate: transcoding: fix encoding_profile leak
Guillaume Desmottes [Tue, 24 May 2016 14:05:24 +0000 (14:05 +0000)]
validate: transcoding: fix encoding_profile leak

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D980

8 years agovalidate: transcoding: fix caps leaks
Guillaume Desmottes [Tue, 24 May 2016 14:05:17 +0000 (14:05 +0000)]
validate: transcoding: fix caps leaks

The 'all_raw_caps' list is never used and was just leaking caps.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D979

8 years agovalidate: scenario: always unref srcpad
Guillaume Desmottes [Tue, 24 May 2016 14:05:10 +0000 (14:05 +0000)]
validate: scenario: always unref srcpad

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D959

8 years agovalidate: media-descriptor-writer: fix pad leaks
Guillaume Desmottes [Tue, 24 May 2016 14:05:03 +0000 (14:05 +0000)]
validate: media-descriptor-writer: fix pad leaks

- the pad returned by gst_element_get_static_pad() was leaked.
- unref the pad from snode when updating it, not the pad passed as
callback to pad_added_cb()

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D958

8 years agovalidate: clean up action_types list on deinit
Guillaume Desmottes [Tue, 24 May 2016 14:04:57 +0000 (14:04 +0000)]
validate: clean up action_types list on deinit

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D957

8 years agovalidate: keep a ref when registering an override
Guillaume Desmottes [Tue, 24 May 2016 14:04:50 +0000 (14:04 +0000)]
validate: keep a ref when registering an override

_add_override_from_struct() could, in theory, register more than once
the same override so we should not transfer the ref.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D956

8 years agovalidate: use g_object_unref() on GstValidateOverride
Guillaume Desmottes [Tue, 24 May 2016 14:04:44 +0000 (14:04 +0000)]
validate: use g_object_unref() on GstValidateOverride

Those are GObject subclasses, not GstObject.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D955

8 years agovalidate: add mesa-related valgrind suppressions
Guillaume Desmottes [Tue, 24 May 2016 14:04:37 +0000 (14:04 +0000)]
validate: add mesa-related valgrind suppressions

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D954

8 years agovalidate: generate valgrind suppression traces
Guillaume Desmottes [Tue, 24 May 2016 14:04:31 +0000 (14:04 +0000)]
validate: generate valgrind suppression traces

Makes fixing easier as then we can just re-use the generated trace.

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D953

8 years agovalidate: fix typo if Xvfb is not installed
Guillaume Desmottes [Tue, 24 May 2016 14:04:25 +0000 (14:04 +0000)]
validate: fix typo if Xvfb is not installed

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D375

8 years agopad-monitor: Don't check for DISCONT flags on buffer in pull-mode
Edward Hervey [Mon, 23 May 2016 13:57:04 +0000 (15:57 +0200)]
pad-monitor: Don't check for DISCONT flags on buffer in pull-mode

It only makes sense in push-mode

8 years agopad-monitor: Update checks for MISSING_DISCONT
Edward Hervey [Fri, 20 May 2016 07:07:01 +0000 (09:07 +0200)]
pad-monitor: Update checks for MISSING_DISCONT

* Some SEGMENT might be updates caused by calling gst_pad_set_offset(),
  which will send the same segment but with an updated offset and/or
  based field. For those segments, we don't require a DISCONT on the
  following buffer.
* Ignore differences in flags, they aren't relevant for now to figure
  out whether the segment is an update or not
* Ignore difference in 'position', it's only meant for internal usage
  by elements.
* Changes in the end position (stop in forward playback and start in
  reverse playback) are considering updates

Furthermore, also expect a DISCONT flag on the first buffer following
a STREAM_START.

8 years agogst-validate: Avoid overreading array
Edward Hervey [Tue, 17 May 2016 08:03:26 +0000 (10:03 +0200)]
gst-validate: Avoid overreading array

If we offset the argv table as argument, we need to decrement the number
of items in the array (argc) when iterating it

8 years agopad_monitor: Add a check for buffer DISCONT flag
Edward Hervey [Thu, 19 May 2016 09:59:19 +0000 (11:59 +0200)]
pad_monitor: Add a check for buffer DISCONT flag

The first buffer after a FLUSH or SEGMENT should have the DISCONT flag
set.

8 years agolauncher: set gsettings-backend to prevent weird deadlocks
Thiago Santos [Sun, 15 May 2016 09:34:21 +0000 (06:34 -0300)]
launcher: set gsettings-backend to prevent weird deadlocks

It seems like some sort of forking/dconf/gtype combination can
deadlock occasionally. Setting the gsettings backend to memory
makes it go away.

Same issue: https://cgit.freedesktop.org/gstreamer/gst-plugins-good/commit/tests/check/Makefile.am?id=8e2c1d1de56bddbff22170f8b17473882e0e63f9

8 years agovalidate: add non-seekable fragment file seek tests to the blacklist
Thiago Santos [Sat, 7 May 2016 00:27:53 +0000 (21:27 -0300)]
validate: add non-seekable fragment file seek tests to the blacklist

The file has no index and is not easily seekable, seeking in these
kind of files isn't implemented.

8 years agodebug-viewer: Allow running uninstalled with symlink
Nicolas Dufresne [Tue, 3 May 2016 19:47:32 +0000 (15:47 -0400)]
debug-viewer: Allow running uninstalled with symlink

When uninstalled, we look at the directory of the executable to find
the resources. This patch uses realpath in replacement to abspath so
the path get expended, and symlink are followed.

8 years agovalidate: Stop using deprecated method
Alexandru Băluț [Wed, 13 Apr 2016 19:18:28 +0000 (21:18 +0200)]
validate: Stop using deprecated method

https://bugzilla.gnome.org/show_bug.cgi?id=764964

8 years agoscenarios: Fix scrub_backward scenarios
Edward Hervey [Thu, 7 Apr 2016 12:11:45 +0000 (14:11 +0200)]
scenarios: Fix scrub_backward scenarios

We will be changing states, set the handles-state variable accordingly

8 years agoAdded menu opts to filter in instead of only out
Xabier Rodriguez Calvar [Fri, 18 Mar 2016 09:42:55 +0000 (10:42 +0100)]
Added menu opts to filter in instead of only out

Added also menu option to filter a log level and all above that

https://bugzilla.gnome.org/show_bug.cgi?id=763857

8 years agoAdded support to filter in instead of only out
Xabier Rodriguez Calvar [Fri, 18 Mar 2016 09:42:18 +0000 (10:42 +0100)]
Added support to filter in instead of only out

Added support to filter a log level and all above it

https://bugzilla.gnome.org/show_bug.cgi?id=763857

8 years agofilter: add more logging
Stefan Sauer [Mon, 4 Apr 2016 20:23:26 +0000 (22:23 +0200)]
filter: add more logging

8 years agomenu: port menu.popup calls
Stefan Sauer [Fri, 25 Mar 2016 21:20:11 +0000 (22:20 +0100)]
menu: port menu.popup calls

Add 1 extra arg.

8 years agodebug-viewer: use the gi GLib version rather than the old gobject glib package
Philippe Normand [Thu, 14 Jan 2016 11:21:34 +0000 (12:21 +0100)]
debug-viewer: use the gi GLib version rather than the old gobject glib package

8 years agodebug-viewer: explicitely require GTK+3
Philippe Normand [Thu, 14 Jan 2016 11:19:22 +0000 (12:19 +0100)]
debug-viewer: explicitely require GTK+3

8 years agodebug-viewer: add more logging and some profiling how-to
Stefan Sauer [Tue, 12 Jan 2016 08:44:17 +0000 (09:44 +0100)]
debug-viewer: add more logging and some profiling how-to

The port is much slower than the gtk3 version. Try to figure why.

8 years agoGstDebugViewer/Timeline: port timeline widgets to gtk3
Stefan Sauer [Wed, 21 Oct 2015 13:03:03 +0000 (15:03 +0200)]
GstDebugViewer/Timeline: port timeline widgets to gtk3

8 years agoGstDebugViewer: basic port to gtk3 and python gobject
Stefan Sauer [Tue, 20 Oct 2015 13:21:01 +0000 (15:21 +0200)]
GstDebugViewer: basic port to gtk3 and python gobject

8 years agogst-validate: Link with GIO and clean up CFLAGS/LIBADD/LDFLAGS
Tom Schoonjans [Fri, 25 Mar 2016 14:17:28 +0000 (14:17 +0000)]
gst-validate: Link with GIO and clean up CFLAGS/LIBADD/LDFLAGS

https://bugzilla.gnome.org/show_bug.cgi?id=764192

8 years agovalidate: use new gst_element_class_add_static_pad_template()
Vineeth T M [Sun, 6 Mar 2016 23:55:27 +0000 (08:55 +0900)]
validate: use new gst_element_class_add_static_pad_template()

https://bugzilla.gnome.org/show_bug.cgi?id=763197

8 years agocodecanalyzer: use new gst_element_class_add_static_pad_template()
Vineeth T M [Sun, 6 Mar 2016 23:53:23 +0000 (08:53 +0900)]
codecanalyzer: use new gst_element_class_add_static_pad_template()

https://bugzilla.gnome.org/show_bug.cgi?id=763197

8 years agoBack to development
Sebastian Dröge [Thu, 24 Mar 2016 11:34:03 +0000 (13:34 +0200)]
Back to development

8 years agoRelease 1.8.0
Sebastian Dröge [Thu, 24 Mar 2016 11:11:29 +0000 (13:11 +0200)]
Release 1.8.0

8 years agovalidate: launcher: Make sure to properly setup all testsuites
Thibault Saunier [Wed, 23 Mar 2016 19:02:47 +0000 (20:02 +0100)]
validate: launcher: Make sure to properly setup all testsuites

When a first testsuite will set paths, it does not mean that we should
just register following testsuite test manager default tests.

So we need to make a difference between the media paths the user passed
with --media-path and the ones defined by the testsuite.

8 years agovalidate:launcher: Handle testslist files even running several testsuites
Thibault Saunier [Wed, 23 Mar 2016 18:34:10 +0000 (19:34 +0100)]
validate:launcher: Handle testslist files even running several testsuites

Only if those testsuites do not use the same Tester as we
currently can't know to what testsuite a test belongs.

8 years agolauncher: Avoid caching all the debug logs in memory
Thibault Saunier [Wed, 23 Mar 2016 10:48:10 +0000 (11:48 +0100)]
launcher: Avoid caching all the debug logs in memory

And just write the temporary XML file on disc

8 years agovalidate: launcher: Blacklist dash and HLS failling seeking tests
Thibault Saunier [Tue, 22 Mar 2016 18:00:48 +0000 (19:00 +0100)]
validate: launcher: Blacklist dash and HLS failling seeking tests

As described in https://bugzilla.gnome.org/show_bug.cgi?id=764020

8 years agovalidate:testsuite: Blacklist validate.hls.playback.scrub_forward_seeking.hls_bibbop...
Thibault Saunier [Tue, 22 Mar 2016 16:46:50 +0000 (17:46 +0100)]
validate:testsuite: Blacklist validate.hls.playback.scrub_forward_seeking.hls_bibbop as it is racy

8 years agovalidate: pad-monitor: Check right segment after seek
Thibault Saunier [Tue, 22 Mar 2016 11:07:08 +0000 (12:07 +0100)]
validate: pad-monitor: Check right segment after seek

After a seek we need to wait for the right segment (meaning the segment
with seqnum == last seek/flush stop seqnum) to check whether the segment.time
has been properly set.

8 years agovalidate: Accept pad return FLUSHING when the element is being teared down
Thibault Saunier [Tue, 22 Mar 2016 10:19:42 +0000 (11:19 +0100)]
validate: Accept pad return FLUSHING when the element is being teared down

In the case and element is in READY or is going to READY state, it can
always return GST_FLOW_FLUSHING.

Avoid a race where a demuxer sinkpad has not been set to FLUSHING when we are
still processing a buffer but downstream is already FLUSHING and thus
the demuxer is already returning FLUSHING.

8 years agovalidate: Use GstValidate logging system to print buffering avancement
Thibault Saunier [Mon, 21 Mar 2016 13:34:27 +0000 (14:34 +0100)]
validate: Use GstValidate logging system to print buffering avancement

So it can be used in the launcher.
And make sure to properly parse the info in the launcher.

8 years agovalidate:launcher: Remove dependency on wget
Thibault Saunier [Wed, 9 Mar 2016 09:05:49 +0000 (10:05 +0100)]
validate:launcher: Remove dependency on wget

8 years agoRelease 1.7.91
Sebastian Dröge [Tue, 15 Mar 2016 10:40:03 +0000 (12:40 +0200)]
Release 1.7.91

8 years agovalidate: Add missing parenthesis to seek position check
Sebastian Dröge [Mon, 14 Mar 2016 12:29:57 +0000 (14:29 +0200)]
validate: Add missing parenthesis to seek position check

https://bugzilla.gnome.org/show_bug.cgi?id=763602

8 years agovalidate: Fix overflow seek position comparision
Sebastian Dröge [Mon, 14 Mar 2016 10:55:57 +0000 (12:55 +0200)]
validate: Fix overflow seek position comparision

MAX(0, ((gint64) priv->segment_start - priv->seek_pos_tol) will be a high
positive number thanks to being interpreted as unsigned values if
segment_start < seek_pos_tol. Fix this by explicitly checking for this case
and only doing the subtraction otherwise.

This fixes the problem from fdccffbb2e5885b3f8e7369cdbda45b6717ffab0
completely now.

https://bugzilla.gnome.org/show_bug.cgi?id=763602

8 years agovalidate: Fix seding mistakes
Thibault Saunier [Tue, 8 Mar 2016 16:01:09 +0000 (17:01 +0100)]
validate: Fix seding mistakes

When we added namespace to make GI happy we ended up with
structure called like GstValidateMediaGstValidateMediaXXX.

8 years agomonitor: Add critical issue for checking accurate seek results
Edward Hervey [Tue, 8 Mar 2016 09:49:43 +0000 (10:49 +0100)]
monitor: Add critical issue for checking accurate seek results

If an accurate seek is accepted, the resulting segment.time should be
exactly the requested seek start value..

https://bugzilla.gnome.org/show_bug.cgi?id=763299

8 years agovalidate: Fix accurate seeking in paused failling condition
Thibault Saunier [Tue, 8 Mar 2016 11:00:19 +0000 (12:00 +0100)]
validate: Fix accurate seeking in paused failling condition

8 years agoValidate: Fix consider, launch spelling mistakes
Vineeth T M [Tue, 8 Mar 2016 06:54:32 +0000 (15:54 +0900)]
Validate: Fix consider, launch spelling mistakes

https://bugzilla.gnome.org/show_bug.cgi?id=763289

8 years agovalidate: Fix wrong condition check when adding tests
Vineeth T M [Mon, 22 Feb 2016 02:49:48 +0000 (11:49 +0900)]
validate: Fix wrong condition check when adding tests

When listing tests, checking whether uri is present or not and displaying error.
But uri does notneed to be present in case of pipeline generator. So the condition check is wrong.
This results in validateelements testsuite not working. Hence modifying the condition to
not error out on valid cases.

https://bugzilla.gnome.org/show_bug.cgi?id=762422

8 years agoRelease 1.7.90
Sebastian Dröge [Tue, 1 Mar 2016 17:23:37 +0000 (19:23 +0200)]
Release 1.7.90

8 years agovalidate:launcher: Strip os.pathsep from extra env variables
Thibault Saunier [Tue, 1 Mar 2016 13:59:29 +0000 (14:59 +0100)]
validate:launcher: Strip os.pathsep from extra env variables

We might be working with something that is not a path

8 years agovalidate: Add a missing new line to media info files
Thibault Saunier [Mon, 29 Feb 2016 14:55:47 +0000 (15:55 +0100)]
validate: Add a missing new line to media info files

8 years agovalidate: Fix build
Thibault Saunier [Tue, 23 Feb 2016 11:15:21 +0000 (12:15 +0100)]
validate: Fix build

8 years agovalidate:launcher: Add a way to fail if test have been removed/added
Thibault Saunier [Tue, 23 Feb 2016 10:29:56 +0000 (11:29 +0100)]
validate:launcher: Add a way to fail if test have been removed/added

8 years agovalidate: Fix annotations
Thibault Saunier [Fri, 22 Jan 2016 19:55:27 +0000 (20:55 +0100)]
validate: Fix annotations

Skipping all functions and methods that are not relevant and not easily
introspectable

Differential Revision: https://phabricator.freedesktop.org/D706

8 years agovalidate: Remove unimplemented function
Thibault Saunier [Fri, 22 Jan 2016 19:45:01 +0000 (20:45 +0100)]
validate: Remove unimplemented function

Differential Revision: https://phabricator.freedesktop.org/D705

8 years agovalidate: Namespace all our structures and objects
Thibault Saunier [Fri, 22 Jan 2016 19:14:16 +0000 (20:14 +0100)]
validate: Namespace all our structures and objects

Making GI a bit happier.

Those are not stable API anyway...

Differential Revision: https://phabricator.freedesktop.org/D704

8 years agovalidate: Misc annotation and gi friendly cleanups
Thibault Saunier [Fri, 22 Jan 2016 18:50:15 +0000 (19:50 +0100)]
validate: Misc annotation and gi friendly cleanups

Differential Revision: https://phabricator.freedesktop.org/D703

8 years agovalidate: Guarantee that we never create 2 monitors for the same object
Thibault Saunier [Fri, 22 Jan 2016 18:38:53 +0000 (19:38 +0100)]
validate: Guarantee that we never create 2 monitors for the same object

Differential Revision: https://phabricator.freedesktop.org/D702

8 years agovalidate: Turn GstValidateRunner into a GstTracer
Thibault Saunier [Sat, 24 Oct 2015 07:28:51 +0000 (09:28 +0200)]
validate: Turn GstValidateRunner into a GstTracer

This way we do not need the LD_PRELOAD hack anymore

Add a new libgstvalidateplugin GStreamer plugin, making sure it shares
the exact same code as the library (exposing only the wanted symbols).

Fix the way we set where to install GstValidate plugins

Try to keep backward compatibility even if tracers should never be instantiated
after an GstElement has been instantiated.

Differential Revision: https://phabricator.freedesktop.org/D459

8 years agoreporter: Properly reset g_log handler when reporter is destroyed
Thibault Saunier [Fri, 22 Jan 2016 11:58:31 +0000 (12:58 +0100)]
reporter: Properly reset g_log handler when reporter is destroyed

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D699

8 years agovalidate: Make _deinit thread safe
Thibault Saunier [Tue, 19 Jan 2016 10:10:49 +0000 (11:10 +0100)]
validate: Make _deinit thread safe

Reviewed-by: Thibault Saunier <tsaunier@gnome.org>
Differential Revision: https://phabricator.freedesktop.org/D698