va: Delay decoders downstream negotiation.

Delay decoders downstream negotiation just before an output frame
needs to be allocated.

This is required, are least for H.264 and H.265 decoders, since
codec_data might trigger a new sequence before finishing upstream
negotiation, and sink pad caps need to set before setting source pad
caps, particularly to forward HDR fields. The other decoders are
changed too in order to keep the same structure among them.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1257>

vabasedec: Move warning message to decoder's category.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1257>

va: Move common variable need_negotiation to GstBaseDec.

This is a common variable to all decoders, so it's sound to move it to
the base helper.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1257>

va: Move back parent_object to each element.

Using GstBaseDec hack to access the parent_object of each element in
the element itself is a bit fragile. It would be better to keep its
own parent object as the usual global variable. It would make it
resistant to code changes.

The GstBaseDec macro to access the parent object now it's internal to
base decoder.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1257>

video-format: Add support for ARGB64 LE/BE and similar variants

Co-authored-by: Sebastian Dröge <sebastian@centricular.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1247>

identity: Fix crash when receiving a gap event outside the current segment

We were checking if the start time of the gap event was
GST_CLOCK_TIME_NONE, which is superfluous because that cannot happen,
and then not checking if it was NONE after gst_segment_to_running_time,
which caused a crash if an identity received a gap event fully or
partially outside the current segment.

This patch was done in cooperation with:
Jan Alexander Steffens (heftig) <jan.steffens@ltnglobal.com>

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1269>

rtptwcc: don't assume all PacketInfo->data are buffers

They can also be buffer lists

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1268>

neon: Allow building against neon 0.32.x

No API/ABI changes: https://github.com/notroj/neon/blob/0.32.0/NEWS#L3

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1267>

meson: require matching GStreamer dep versions for unstable development releases

Fixes https://gitlab.freedesktop.org/gstreamer/gst-plugins-good/-/issues/929

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1244>

subprojects: fdk-aac: add fallback_url

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1244>

webrtcbin: fix check_negotiation computing on caps event

It seems logical that check_negotiation be true if received_caps
is *not* equal to the new caps.

Also clean up handling of transceivers' ssrc events, as this
patch triggered a leaky code path.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1233>

webrtcbin: connect input stream when receiving caps

.. if a current direction has already been set

When `webrtcbin` has created an offer based on codec_preferences,
it might not have received caps on its sinkpads by the time a
remote description is set, in which case we want to connect the
input stream upon actual reception of the caps instead.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1233>

webrtcbin: consider pads with trans->codec_preferences ready

.. when determining whether we can emit on-negotiation-needed

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1233>

mpeg4videoparse: fix criticals trying to insert configs that don't exist yet

With mpeg4videoparse drop=false config-interval=N|-1 we might be
trying to insert a config before we have actually received one,
in which case we'll try to map a NULL buffer which will generate
lots of criticals.

Fixes #855

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1265>

qtdemux: Fix text and closed-caption handling.

https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1182
broke text and and closed caption extraction when introducing WebVTT
support, by making the output buffers not have timestamps any more.

Fix that by making the process functions copy buffer metadata
when generating new output buffers.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1262>

msdk: Insert hdr sei at hevc encoder

There are two HDR SEIs defined in spec: mastering display colour volume and
content light level. Add insertion of HDR SEIs when they are available
during encoding.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1242>

webrtcbin: Also check data channel transport when collating connection state

https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/838

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1224>

vapostproc: Color fixation will choose othercaps' structure.

gst_va_fixate_format() will iterate all othercaps' structures to find
the one with less information lost at color conversion. If a structure
with same color format is found, the iteration stops. It's like a
smart truncation. Then, this function also will choose the caps
feature.

Later this structure is used fixate its size and no further truncation
is needed.

Don't intersect at fixate, since it kills possible resizing.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1261>

vapostproc: Ignore direction at orientation swapping.

It doesn't matter the direction of the negotiation.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1261>

vapostproc: Consider video orientation for border calculation.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1261>

decodebin3: Only unblock for GAP events and buffers.

An alternate fix for
https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/1239
and https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1236
that makes it clear the intent is to only unblock on a GAP event, and
not any others.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1260>

python: Update audiotestsrc to return correct tuple on FlowReturn Errors

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1259>

avtpsrc: Retry receive with same buffer size

Without this patch in case of a retry recv() will be called with a
negative size argument.

Signed-off-by: Timo Wischer <timo.wischer@de.bosch.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1078>

parsebin: fix critical when sorting pads

If the pad does not have a current caps, get_pad() returns the query
caps which can be ANY. In such case the caps does not have any structure
resulting in a critical warning when calling gst_caps_get_structure().

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1254>

cccombiner: fix default value when installing schedule property

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1252>

cccombiner: fix emission of selected-samples in one case

Detected while reading the code, cccombiner must set
self->current_video_buffer to NULL *after* emitting selected-samples
in order for the application to get a useful return when peeking
the next video sample.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1252>

cccombiner: stop attaching caption buffers when caption pad has gone EOS

When schedule is true (as is the case by default), we insert padding
when no caption data is present in the schedule queue, and previously
weren't checking whether the caption pad had gone EOS, leading to
infinite scheduling of padding after EOS on the caption pad.

Rectify that by adding a "drain" parameter to dequeue_caption()

In addition, update the captions_and_eos test to push valid cc_data
in: without this cccombiner was attaching padding buffers it had
generated itself, and with that patch would now stop attaching
said padding to the second buffer. By pushing valid, non-padding
cc_data we ensure a caption buffer is indeed attached to the first
and second video buffers.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1252>

ci/fuzzing: update for mono repo

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1253>

curlsftpsink: Add support for sha256 fingerprint

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1193>

msdkvpp: Add 12bit formats

Add 12bit formats for different chroma samplings at sink pad and
src pad, including P012_LE, Y212_LE and Y412_LE.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1218>

devicemonitor: g_queue_clear_full introduced in glib 2.60

The GStreamer dependency is glib >=2.56.0. Therefore, define
g_queue_clear_full if glib < 2.60.

Issue added in commit 1912bcbc

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1251>

gs: update README to use fixed versions

Use specific versions, instead of relying on 'master'.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1246>

gssrc: use default blocksize

The blocksize is set to 3 * 1024 * 1024 / 2, which is the default download_size of Google-Cloud-CPP.

Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/846

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1246>

vah265dec: Fix end_picture() vmethod.

Since commit 88437a9c the signature of h265decoder's end_picture()
changed to return GstFlowReturn, but vah265dec was not updated.

This commit fixes this regression.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1248>

devicemonitor: Only fail start() if no provider at all could be started

Also refactor various internals of the monitor code:
  - Don't allow starting twice but just return directly when starting a
    second time.
  - Don't end up in an inconsistent state if call start() a second time
    while the monitor is starting up.
  - Remove complicated cookie code: it was not possible to add/remove
    filters while the monitor was started anyway so this was only useful
    in the very small time-window while starting the monitor or while
    getting the devices. Instead disallow adding/removing filters while
    the monitor is starting, and when getting devices work on a snapshot
    of providers/filters.

Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/667

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1189>

docs: contribute: bump compiler reqs and coding style to C99

Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer-project/-/issues/78

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1243>

typefindfunctions: Use memcmp() instead of a strange macro around strncmp()

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1234>

typefindfunctions: Add various CMAF brands to the MP4 typefinder

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1234>

typefindfunctions: Refactor ftyp brands checking

Store known brands in arrays and add a helper function to check them.

Also add all the `iso3` to `isob` ISO brands and the `hlsf` ISO
fragmented brand.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1234>

typefindfunctions: Add various other boxes to the list of known top-level MP4 boxes

styp, udta, sidx, ssix, prft, mfra, pdin and meta are in the ISOBMFF
spec and emsg in the CMAF spec.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1234>

va: allocator: Fix possible memory leaks

At gst_va_dmabuf_allocator_setup_buffer_full, static code analysis tool
does not know number of objects in descriptor is always larger than 0 if
export_surface_to_dmabuf succeeds. Thus, the tool will assume buf is
allocated with mem but not released when desc.num_objects equals to 0
and raise a mem leak issue.

For gst_va_dambuf_memories_setup, we should also inform the tool that
n_planes will be larger than 0 by checking the value at very beginning.
Then, the defect similar to above will not be raised during static analysis.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1241>

gst-plugins-ugly: update translations

Fixes #656

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1240>

gst-plugins-bad: update translations

Fixes #656

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1240>

gst-plugins-good: update translations

Fixes #656

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1240>

gst-plugins-base: update translations

Fixes #656

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1240>

gstreamer: update translations

Fixes #656

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1240>

d3d11: add support for new debug layer versions

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1237>

qtdemux: Add pasp box to the list of known boxes

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1235>

assrender: Add "application/vnd.ms-opentype" mimetype detection

The "application/vnd.ms-opentype" mimetype is commonly used in many fonts attached in the matroska videos.
Assrender should treat it as compatible without the need of parsing the file extension.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1207>

docs: contribute: update how-to-submit-patches section for monorepo

See https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/840#note_1114907

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1231>

wrap: libopenjp2: use patch version 7

Add support for win32 build
Disable the binary to avoid the thirdparty
dependency to be checked.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1229>

baseparse: fix invalid avg_bitrate after reset

gst_base_parse_reset() does not reset data_bytecount to 0, so
gst_base_parse_update_bitrates() uses a wrong value to calculate
the average bitrate on subsequent pipeline starts. This leads to an
excessive amount of "tag" events being pushed. These events include
very high "bitrate" values that diminish over time, and are produced
until the average bitrate is back to sane values.

Fixes #840

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1228>

Rename toplevel project to gstreamer-full

This generates better tarball names when doing "meson dist" on the top
level project. It is also consistent with that project providing
gstreamer-full library.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1230>

gl/egl: install required egl.h header

Fixes build against using the gir.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1227>

gstvulkan: Constify code in create_shader

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1226>

meson: default to gpl=disabled for gst-plugins-bad and -ugly

This will only affect individual/tarball module builds, as the
options yield to the parent project which was set to gpl=disabled
by default already. We kept it as auto in the original commit
to accommodate the need to update cerbero as well, which had to
be done separately after the initial commit.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1217>

assrender: Do not iterate over mimetypes without filename

No point spending time on iterating and comparing strings if we
are going to reject the value due to missing filename anyway.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1206>

assrender: Fix mimetype detection

Previously gst_structure_has_name was used to get a string to compare with supported mimetypes.
This is incorrect as above function returns a user defined structure name which is
not the structure mimetype value.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1206>

matroska: Set image/attachment structure mimetype

Set structure mimetype to fix data detection by mimetype in other plugins.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1205>

matroska: Treat non-image structure as attachment

Otherwise each structure is named as GstTagImageInfo even if
it does not contain any images which is misleading.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1205>

rtpbin: fix leak of pad when a fec encoder and aux sender a created

The ghost sink pad retrieved by rtpbin from the aux sender was not freed
when there was a previous element (fec encoder) in the chain.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1222>

ci: gstreamer-full: remove static build

The gstreamer-full jobs act as
static builds,  no need to perform it twice.

gstreamer-full-minimal can be nodebug.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1215>

README: Remove outdated Windows ARM64 related comment

We bumped required meson version which contains the mentioned
meson patch already.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1221>

vulkan/swapper: add some debug logging for surface size and present modes

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1219>

interlace: Replace custom lock with object lock

The object lock is sufficient for the task of protecting against
object property data races.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1039>

interlace: Protect all properties with the lock

Avoid blatant data races here.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1039>

interlace: Reset src_fps_d together with src_fps_n

These fields belong together.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1039>

interlace: Clear stored_fields together with stored_frame

These fields belong together.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1039>

interlace: Reset after changing state to READY

Trying to reset before the pads have been deactivated races with the
streaming thread. There was also a buggy buffer clear leaving a dangling
`stored_frame` pointer around. Use `gst_interlace_reset` so this happens
properly.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1039>

nvh264enc: add constrained-baseline to the caps profiles

In practice, when baseline is requested from the encoder it
produces constrained baseline, and it is already reflected
in the profile-iop flags.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1213>

mediafoundation: Fix for UWP build

We don't support D3D11 interop for UWP because some APIs
(specifically MFTEnum2) are desktop application only.
However, the code for symbol loading is commonly used by both UWP and WIN32.
Just link GModule unconditionally which is UWP compatible, and simply don't
try to load any library/symbol dynamically when D3D11 interop is unavailable.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1216>

Revert "ci: Split base and core valgrind on different runners"

Those job are not the ones that take most time in normal run (they
where when running with --n-runs) and that leads to using more runners
which might be a problem at this point.

This reverts commit 4ad22b05a15056b9e85468af266bacc7d3758ee9.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1212>

docs: remove mms plugin from symbol index

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1211>

mms: remove mmssrc plugin

Doubtful that anyone still needs that or there are even
any streams left out there.

MMS was deprecated in 2003 (in favour of RTSP) and support for
it was dropped with Microsoft Media Services 2008.

https://en.wikipedia.org/wiki/Microsoft_Media_Server
https://sdp.ppona.com/news2008.html

Fixes https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/issues/821

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1211>

streamsynchronizer: set running time offset on events

It's cleaner and more generic than overriding the qos events.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1210>

concat: fix qos event handling

We were shadowing the sinkpad variable resulting in:
- the QoS event to be send to all sink pads instead of the active one
- the pad to be leaked

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1208>

doc: update meson version in README.md

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1209>

meson: va, msdk: simplify dep.get_variable() use

With recent Meson versions we can just write dep.get_variable('foo')
instead of dep.get_variable(pkgconfig: 'driverdir', internal: 'driverdir').

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1183>

meson: fix use of deprecated meson api external_program.path()

Just using .full_path() instead.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1183>

meson: fix use of deprecated meson api meson.has_exe_wrapper()

Use meson.can_run_host_binaries() instead.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1183>

meson: update for meson.build_root() and .build_source() deprecation

-> use meson.project_build_root() or .global_build_root() instead.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1183>

meson: update for dep.get_pkgconfig_variable() deprecation

... in favour of dep.get_variable('foo', ..) which in some
cases allows for further cleanups in future since we can
extract variables from pkg-config dependencies as well as
internal dependencies using this mechanism.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1183>

meson: clean up conditional paths after version bump

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1183>

assrender: fix smooth scaling by disabling hinting

When ass hinting value is set to anything other than NONE,
subtitles cannot use smooth scaling, thus all animations will jitter.

The libass author warns about possibility of breaking some scripts when it is enabled,
so lets do what is recommended and disable it to get the smooth scaling working.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1201>

discoverer: Set number to stream infos

The idea is that we can reference to streams using this unique number,
within the context of that discoverer info. That number should always
be usable to reference the streams for a specific stream.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/897>

discoverer: Don't ref NULL taglist

Fixes warning introduced in 064f7bbbfac9610fd391c45091e35391b80e1212

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1200>

integration/ges: Ignore GstStream for validate tests

Requires the ges runner to be configured in such a way that the resulting
stream-id is consistent (and therefore can be checked)

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1187>

stream: Set the object name to the stream id

Makes it more meaningful in debug logs (instead of streamNNNNN)

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1187>

docs: remove mention of mms from docs and mmssrc from tutorials

Prerequisite for our removing it from the cerbero build, see
https://gitlab.freedesktop.org/gstreamer/cerbero/-/merge_requests/760

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1199>

rtphdrext-clientaudiolevel: Rename RFC 6464 element

Multiplying elements named after RFC numbers is confusing,
so let's give them meaningful names.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1125>

ci: gstreamer-full: reenable build

The gstreamer-full job should be run
during the integrate stage always.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1190>

subprojects: Update ogg and vorbis wraps

... to 1.3.5-2 and 1.3.7-2, respectively. Specifically to fix
ogg/vorbis plugin loading on Windows devenv environment.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1196>

Revert "meson: Link to objects instea of static helper library"

This reverts commit b19de413b94d228b1460b0899f9b41b2b5233943.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1194>

Revert "audio: Merge simd libs into the main one"

This reverts commit 4d3a200358439e6c76c5dd1be2daae0cceb0e44a.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1194>

Add initial meson devenv support

There is still a few things missing compared to gst-env.py, but it is
functional.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1162>

video-overlay-composition: Fix constructor guard to actually allow NULL as parameter

And don't use a NULL rectangle later.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1195>

gl/eagl: fix eagl display creation

The GstGLDisplayType of the display was win32 instead of the required eagl.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1188>

qtdemux: Add support for wvtt (WebVTT) subtitles.

WebVTT in ISO MP4 is specified in ISO 14496-30,
and needed for DASH support. It's stored in an
mp4 specific format. To handle it compatibly,
the wvtt boxes are converted back into WebVTT text
and pushed as application/x-subtitle-vtt

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1182>

isomp4: Split buffer process functions.

Split the different handling for closed captions, VobSub subpicture
and timed text samples into separate simplified process functions.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1182>

isomp4: Use a function pointer for buffer splitting.

Swap the `need_process` boolean check on qtdemux streams
for a direct function pointer to the splitting function,
so we can stop adding extra cases to the single growing
`gst_qtdemux_process_buffer()` function.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1182>