- check the private u64 value rather than fd (gdb problems)
- move the run call outside the connect routine (locking issues)
- try to fix more leaks (PR_Cleanup)
Rather than check for specific glibc version on PLATFORM_LINUX, check
the existance of qsort_r.
This fixes the following compile error with musl libc:
CC src/main.o
In file included from src/main.c:39:0:
src/tvheadend.h:608:20: error: missing binary operator before token "("
#if !__GLIBC_PREREQ(2,8)
^
this is conditional based on a test run from the configure script, as this
type of test is certainly not 100% cross platform compatible.
However its very useful to have a true check of lock ownership rather than
the rather arbitary check that "someone" holds the lock.
Heading towards the start of a tsfile based system that will be
both a useful debugging tool and also a useful starting point for
the generic mpegts framework
Note: this can be compile switch disabled (./configure --disable-trace)
Note: to enable subsystem output use --trace X,Y,Z etc... (--trace all
will enable the lot).
Note: the only way to know what subsystems are valie is to read the code!
Note: --trace implies -d
Should a recording be moved (within same dir) the DVR DB will be
updated.
Should it be moved (out of directory) or deleted entirely, it will
simply be marked as missing.
The parent directories are also monitored should they be moved
or deleted.
Currently this supports pause/resume, and speed control. FF up to 4x uses
full frame output, faster than that or reverse uses i-frame only output.
This causes problems with some players and needs work.
Also buffers are done at the subscription level which means the disk space
is not shared even if it holds the same content. And more importantly
this means you cannot yet record the timeshift buffer like on a standard
PVR.
This allows file:// paths to be specified for channel icons even if image
cache support is disabled.
The image cache functionality is compile time optional (for those without
curl support) and also run-time configurable for those that don't want it.
All images, including EPG ones should be cached.
I have decided to include this as there is some suggestion the performance
will be better on ARM (non-x86) processors where we currently have no
optimised code.
