Commit graph

9862 commits

Author SHA1 Message Date
Wim Taymans
ac1e923617 design: add some ideas for SKIP mode trickmodes 2009-10-27 15:47:39 +01:00
Wim Taymans
03baa04059 multiqueue: buffering is implemented now 2009-10-27 14:40:19 +01:00
Wim Taymans
b852da60d4 multiqueue: make sure percent increases
Keep track of the last posted percent message and make sure the next percent
messages are strictly increasing.
2009-10-27 14:40:17 +01:00
Wim Taymans
30fc865162 multiqueue: refactor buffering code
Move the buffering update code to a separate function so that we can call it
when the buffering state changes due to EOS.
Avoid dividing by 0.
2009-10-27 14:40:15 +01:00
Wim Taymans
99dd9830a9 multiqueue: add buffering support
Add support for buffering mode where we post BUFFERING messages based on the
level of the queues. It currently operates on the first queue that goes over or
under the high/low thresholds.
2009-10-27 14:40:12 +01:00
Wim Taymans
07991ae4ab multiqueue: don't check visible items in buffering
In buffering mode we want to ignore the max visible items to decide when the
queue is filled. Instead, we only look at the number of bytes and/or time in the
queue.
2009-10-27 14:40:09 +01:00
Wim Taymans
6ac9191527 multiqueue: hook up low/high percent
Hook up the low/high percent properties for the buffering mode.
2009-10-27 14:40:07 +01:00
Wim Taymans
b9d4f6e550 multiqueue: hook up property for buffering 2009-10-27 14:40:02 +01:00
Wim Taymans
c6b0d4a43d multiqueue: small cleanups
Remove unused variable to avoid confusion
Fix some typo
2009-10-27 13:22:47 +01:00
Wim Taymans
6082b0bc89 multiqueue: add FIXME for wrong code
Needs further investigation
2009-10-27 13:18:31 +01:00
Wim Taymans
354cf53f81 multiqueue: fix debug output 2009-10-27 13:18:29 +01:00
Wim Taymans
606852c586 multiqueue: avoid shadowing function argument
Don't shadow the sq argument in the underrun_cb function but use
a different variable name to iterate the other queues.
Use the same variable name in the overrun_cb function.
2009-10-27 13:18:25 +01:00
Wim Taymans
814ee012e7 multiqueue: make queue arg explicit
Make the queue argument to IS_FILLED explicit
2009-10-27 13:18:20 +01:00
Wim Taymans
c490d7542c multiqueue: fix properties
Fix properties, make the extra-size properties as not implemented.
2009-10-27 13:18:00 +01:00
Stefan Kost
b83fb45fac debugutils: allow to hide/show pad status with graphdetails flag 2009-10-27 12:06:35 +02:00
Sebastian Dröge
c57e5bf07e typefindhelper: Remove obsolete FIXME
It's not necessary (and not a good idea) to cache the typefind factory
list anymore.
2009-10-24 13:14:25 +02:00
Sebastian Dröge
b04f7489bf registry: private is a C++ keyword, don't use it
...otherwise C++ compilers will complain when including gstregistry.h
2009-10-24 11:58:25 +02:00
Tim-Philipp Müller
4a9e983f4a docs: add Since tag to docs for new API
And tell gtk-doc that GstRegistryPrivate is private.
2009-10-24 10:31:08 +01:00
Edward Hervey
5067664c09 typefind: Keep typefind factories sorted in the registry. Fixes #599147
This avoids having to do the sorting everytime we use typefind

The behaviour of gst_type_find_factory_get_list has subtlely changed
in the sense that the order was previously undefined, whereas now
it returns them sorted by rank and then by name.
2009-10-24 10:44:09 +02:00
Edward Hervey
c79ed99bab registry: Cache element and typefind factories. Fixes 598896
This avoids unneeded list/filtering if the registry hasn't changed
2009-10-24 10:44:04 +02:00
Edward Hervey
ebee258806 gstpluginfeature: API : new gst_plugin_feature_list_copy() method
This allows copying AND incrementing the refcount at the same time,
avoiding a double iteratio of the GList
2009-10-24 10:39:30 +02:00
Edward Hervey
9e792ee5b8 gstregistry: Add a cookie for detecting feature list changes
We also create a private structure, since we will need to add more
data there in following patches.
2009-10-24 10:39:13 +02:00
Wim Taymans
db79dd88c3 message: don't use typechecking cast macros
Simply use casting macros for accessing the message fields like we do for
buffers and events. Avoids some costly typechecking that does not really buy us
much.
2009-10-23 13:40:32 -04:00
Wim Taymans
c9a7e2d188 Revert "gstmessage: Avoid expensive src/type/timestamp fetch."
This reverts commit 61cf93a334.
2009-10-23 13:13:52 -04:00
Edward Hervey
61cf93a334 gstmessage: Avoid expensive src/type/timestamp fetch.
If we've already checked that we have a valid message, use the entries
directly.
2009-10-23 18:21:47 +02:00
Edward Hervey
e5a13bd3fc gstcaps: Use inlined version of _is_any()/_is_empty()
CAPS_IS_ANY and CAPS_IS_EMPTY are the equivalent of their gst_caps_*
counterpart except that they avoid the typechecking and are inlined.

CAPS_IS_EMPTY_SIMPLE only checks for empty caps (without checking if
the caps is ANY).
2009-10-23 17:47:43 +02:00
Wim Taymans
c10690aea2 TODO: add item to TODO list
We currently do a little too much work when we push the first buffer around
resulting in excessive caps checking. We can probably make this a little less
expensive.
2009-10-22 16:42:13 -04:00
Jan Schmidt
dc2c4de73a gstpipeline: Simplify base time checking slightly
Simplify checking and distribution of the base time - don't re-check
the value of a local variable that was set 3 lines earlier.
2009-10-22 12:54:17 +01:00
Edward Hervey
517ab2860e gsturi: Optimisation: Avoid type-checking in sorting method.
We already know the list only contains plugin features
2009-10-22 13:16:07 +02:00
Edward Hervey
a9ed7c0ec4 gsturi: Optimisation : Cast when we're sure of the type.
Also directly access GstElementFactory->uri_type instead of going
through a function that will (once again) check whether it's a
GstElementFactory
2009-10-22 13:13:56 +02:00
Edward Hervey
5583220a1d tests/fakesink: Add some debugging 2009-10-22 12:33:37 +02:00
Edward Hervey
ca054ff4e5 tests/faeksink: Lower the number of threads to avoid timeouts
We just end up with way too much contention in glib otherwise.
2009-10-22 12:33:01 +02:00
Edward Hervey
4b13520b61 gstregistry: Use hash table when finding a feature. Fixes #598895 2009-10-22 11:13:47 +02:00
Wim Taymans
2f7b348fc0 tee: implement custom acceptcaps function
Implement a custom acceptcaps function on the sinkpad. We can accept any caps as
long as it is accepted by all downstream peer elements.
2009-10-21 16:26:01 -04:00
Wim Taymans
b95cc5031b multiqueue: avoid lock for taking the counter
The counter for incomming data is already protected with the STREAM_LOCK so we
don't need to add another lock around it.
2009-10-21 13:38:57 -04:00
Wim Taymans
368b7899c0 registry: hash the plugin basename
Maintain a hashtable of the plugin basename. We can then use this
hashtable to speedup the search for an existing plugin and avoid
a whole lot of strcmp calls.
2009-10-20 23:28:54 -04:00
Wim Taymans
5de94ede2f registry: speed up _strlen
Make the _strlen function a little tighter
2009-10-20 23:27:41 -04:00
Wim Taymans
50b9a3ecc3 registry: name is never NULL
When looking up a feature by name, we never call this internal
function with NULL so we don't have to check for it.
2009-10-20 21:43:58 -04:00
Wim Taymans
d722d16155 registry: refactor plugin lookup
We keep lookup plugins by their basename. Avoid creating a basename
from a filename if we can.
2009-10-20 21:39:11 -04:00
Wim Taymans
1be3d5cd83 registry: do quick check for . files
Do a quick check for . files before calling the strcmp functions
2009-10-20 21:15:56 -04:00
Michael Smith
d833b39a85 Remove executable bits on xml unit test. 2009-10-20 12:21:41 -07:00
Benjamin Otte
868711fc70 docs: Fix docs for gst_caps_set_simple() 2009-10-20 10:51:25 +01:00
Jan Schmidt
08c56edb84 docs: Modify docs string slightly.
Apparently starting the last line of a docs string with 'returns' both
confuses and enrages gtk-doc. Use a slightly different wording instead.
2009-10-19 13:02:30 +01:00
Jan Schmidt
de199b4d8d pluginloader: When a plugin is blacklisted, output a GST_ERROR line. 2009-10-19 12:30:07 +01:00
Edward Hervey
adfb63837c tools/gst-inspect: Check we're not handling NULL pointers. 2009-10-19 13:30:10 +02:00
Edward Hervey
b11d72adb4 tools/gst-inspect: Remove dead assignment 2009-10-19 13:29:40 +02:00
Peter Kjellerstedt
485ae69c9e trace: Do not poison gst_trace_add_entry()
Since gst_trace_add_entry() is a macro, gcc will barf when it is
defined in case it has been poisoned due to trace support being
disabled.
2009-10-19 10:53:00 +02:00
Stefan Kost
382ad3baa6 controller: just cast in internal API where we have checked parameters already 2009-10-18 23:18:58 +03:00
Stefan Kost
a4e4916632 controller: use g_slice for controlled property structures
Use g_slide instead of nomal g_new, Also don't init struct with 0 as we need to
init it anyway with the real values.
Also join the 3 flags checks into one.
2009-10-18 23:15:07 +03:00
Edward Hervey
bb0918279f libs/base/typefindhelper: Remove useless typechecking in tight loop
The list against which we run the comparefunc will only contain
GstPluginFeature, therefore remove the 6 expensive type checks we do
for every single comparision.
2009-10-18 17:17:17 +02:00