* Renamed the "Option" menu to "Show", removed the now redundant "Show" from
the items in that menu.
* Add an "Only featured packages" item in the Show menu. This item is disabled
as soon as the user types something in the search terms field. Othetwise,
this setting is respected, i.e. one can combine the featured setting with
other settings, i.e. show all featured packages of a certain depot and in a
certain category.
* Added PackageInfo::IsProminent() and removed the respective method from
the MainWindow (which only considered a package prominent if it was not
yet installed).
* "Show -> Installed packages" now defaults to on. There have been complaints
on Trac and I guess rightfully so. The use-cases in which one starts
HaikuDepot and wants to know about installed packages are just as likely
as the use-case where one wants to install new software.
Don't replace the ref with itself in the most common case where the
scanned file is not a symlink. Save a lot of useless memory copies and
ref flattening in BMessage.
These were outsourced and are not Haiku dependencies anymore.
This list may need some more cleanup: I'm not sure why we credit Bullet
Physics, for example.
The ffmpeg encoders are restricted on which raw formats they accept
(each with different constraints). When asked to encode something, the
ffmpeg encoder now checks the codec capabilities and if the suggested
input format does not match, modifies it to something the codec can
actually handle.
Enable "experimental" codecs, as this is needed for AAC support in ffmpeg 0.10.
This allows format negociation to work properly, and as a result we can
enable several new audio formats: AAC (for MPEG family only), AC3, and
raw audio. Declare MKV as B_ANY_FORMAT_FAMILY since it is designed to be
a generic container (so it is possible to put AAC in MKV). Also remove
duplicate entries for AC3.
The packages count view centers its label. It therefore needs a full update
on resize. Also optimize SetItemCount() by trying to avoid unneccessary
layout invalidations.
This would lead to an attempt to read beyond the track end, making
MediaTrack return an error, and finally causing MediaConverter to not
flush the end of the encoded file to disk properly.
BuildReportList checks that the array does not already holds the reports
before adding them. If one of the uninitialized elements happened to be
the address of one of the reports, it would not get added again,
possibly leading to items in the array being in the wrong order. Just be
safe and clear the array.
Fixes#10221, patch there not used as it cluttered the code more than
needed.
* This was never implemented and no one noticed until now.
* A default value for a pointer doesn't make that much sense anyway, so
using the FindPointer method is fine.
* synth_file_reader: TList to BObjectList.
* delete two old prototypes of MidiPlayer and the README noting that
they should be removed
* delete reference to PatchBay which was moved to src/apps in 2013.
Fixes#7027.
BeOS did allow other threads than the owner to unlock a BLocker (the be
book says so). We did not, and silently ignored the unlock attempt in
this case, probably resulting in a deadlock of applications using the
feature.
Call debugger instead so:
* The problem is made visible for such apps
* The debugger call is continuable so the app can be run, still
Will help making a decision on what to do here (follow BeOS or change
behavior) and make a final decision for #6400.
int64_t is signed. Although it does not make a difference by itself, because
INT64_MAX is still a valid number for uint64_t (UL), the later INT64_MIN
declaration depends on INT64_MAX, and therefore got implicitly casted to
unsigned type.
This fixes the following program on a x86_64 system:
#include <stdint.h>
int main() {
int64_t test = 5;
if (test < INT64_MIN)
return 1;
return 0;
}
This is a regression since commit 1d13a609 ("stdint.h: define [U]INT64[MAX|MIN]
with [U]L on x86_64").
Signed-off-by: Jerome Duval <jerome.duval@gmail.com>
