Darn, and here I was hoping there was a policy change in favor of
vector intrinsics ;-)
Would a patch replacing some of my mmx.h-based work with vector
intrinsics be accepted if the mmx.h macro code were retained and
selected via an #ifdef for older compilers?
On Tue, 28 Dec 2004 19:19:57 -0500,
Wendy Seltzer wrote:
I think I've tracked down the problem here but don't have the C++
know-how to fix it:
Reading the track length of flac files appears to have broken in the big
Nov. 26 mythmusic patch. A rollback to CVS from Nov. 25 reads track
lengths again (and once they're in the db,
I have just found a problem (no solution yet) with
a dvbt recorded program.
happens about 40min into the file.
file is nuv transcoded so mpeg4.
a partial trace of the interesting bit
is
2004-12-30 00:07:37.029 _AddSamples bytes=9216,
used=161793, free=350207, timecode=954437762004-12-30
I finally discovered why I cannot record three HD shows while watching a
fourth, for more than an hour.
In the DTV processing of keyframes, it was calling SetPositionMap
instead of SetPositionMapDelta. This means that it was trying to
delete/insert the position map data for the *entire* show,
Doug Larrick wrote:
libavformat/libavcodec expose none of this except number of channels,
and even that is not properly initialized (always 2) by the time we
use it in avformatdecoder.cpp. My immediate problem would be solved
if I fixed this bug, because it turns out the DVS stream is mono.
Daniel Thor Kristjansson wrote:
Most of audio information with ATSC is not in the stream that ffmpeg
sees. Even the limited audio descriptor sometimes present in the PMT is
jettisoned with the PMT rewriting. I don't think ffmpeg even tries to
look at this information, but we don't save it in
On Wed, 29 Dec 2004, Doug Larrick wrote:
]Daniel Thor Kristjansson wrote:
] Most of audio information with ATSC is not in the stream that ffmpeg sees.
] Even the limited audio descriptor sometimes present in the PMT is jettisoned
] with the PMT rewriting. I don't think ffmpeg even tries to look at
Hi,
When using Browse mode while watching live tv, if you pass a channel for
which there is no current program data, junk is displayed instead of the
channel name/number.
This patch makes ProgramInfo::GetProgramAtDateTime() check if
ProgramList::FromProgram() returned any results, and if not
On Wed, 29 Dec 2004, Daniel Thor Kristjansson wrote:
On Wed, 29 Dec 2004, John Patrick Poet wrote:
]I finally discovered why I cannot record three HD shows while watching a
]fourth, for more than an hour.
]In the DTV processing of keyframes, it was calling SetPositionMap instead of
No, I only have Intel processors at my disposal. Compiling for pentium4
gives me a 30% speed boost. For the rest, I based this on the gcc
manual. I've heard rumors about xp optimization being broken, but I
can't verify this myself.
Not really an authority at all but this article does touch on
There are two patches here. The first, cleanup-verbose-nvr-v1.patch,
replaces most of the cerr redirects with VERBOSE macros. This one should
be completely safe to apply.
The second, cleanup-exit-nvr-v1.patch, replaces a number of exits with a
errored boolean and some tests of that variable.
On Wed, 29 Dec 2004, Jason Gabriele wrote:
]Not really an authority at all but this article does touch on this:
]http://www.anandtech.com/linux/showdoc.aspx?i=2308p=10
]
]However, this is just gzip compression. I would like to see a shootout
]on video compression/playback.
Yeah, this is
While a ringbuffer is still necessary to prevent the overruns, this fix does
mean that the ringbuffer size can be much smaller. In a preliminary test, I
saw a maximum of 4.7MB used after an hour of recording three shows. These
shows were just upconverts and not true HD, but that is
Jack Porter schreef:
Hi,
When using Browse mode while watching live tv, if you pass a channel
for which there is no current program data, junk is displayed instead
of the channel name/number.
This patch makes ProgramInfo::GetProgramAtDateTime() check if
ProgramList::FromProgram() returned any
I just noticed this error in my mythbackend.log. This is a fresh
build from cvs. Any ideas as to what this is?
thanks,
Pb
--
Utque fit, in gremium pulvis si forte puellae
Deciderit, digitis excutiendus erit:
Etsi nullus erit pulvis, tamen excute nullum:
Quaelibet officio causa sit apta
On Wed, 29 Dec 2004, Taylor Jacob wrote:
While a ringbuffer is still necessary to prevent the overruns, this fix does
mean that the ringbuffer size can be much smaller. In a preliminary test, I
saw a maximum of 4.7MB used after an hour of recording three shows. These
shows were just
The upconverted shows are not stretched. They are pillar boxed. That much
black area should be easily compressed, so I assumed that the station would
dial down the bitrate
The concept of upconverting is broken in itself, and from the example I have
seen here the bitrate is fixed
At 11:15 AM 12/29/2004 +, Colin Guthrie wrote:
The attached patch contains both the length fix and the insertion of these
#defines into three files. Feel free not to apply the #define bits if it's
not appropriate (should be trivial to strip out of the patch).
Thanks for the quick response
Wendy Seltzer wrote:
At 11:15 AM 12/29/2004 +, Colin Guthrie wrote:
The attached patch contains both the length fix and the insertion of
these #defines into three files. Feel free not to apply the #define
bits if it's not appropriate (should be trivial to strip out of the
patch).
Thanks
I don't know if it's related or not, but there is some timecode wierdness in
mythtv.
A while back there was some discussion about the audio time code sync
packets being wrong for some reason and that caused playback problems.
If i remember correctly the source of the problem was not found and
This patch cleans up the handling of errors encountered when executing
an external channel changing program or script. I've also added some
code to timeout if this program does not exit within 30 seconds. However
I've left in the in the execution of the script because this
function is not
This removes exit()'s in datadirect.cpp, and changes grabLineupsOnly(),
grabData() and grabAllData() so that they return true on successful
completion, and false on unsuccessful completion. It also adds checks
for error conditions, where appropriate.
-- DanielIndex: libs/libmythtv/datadirect.h
On Saturday 25 December 2004 04:19 pm, Shane Shrybman wrote:
Hi,
This is almost definitely not the correct fix for this. However, I hope
it does illustrate the problem.
This bug causes the SBE to crash when the MBE is restarted. This patch
allows my SBE to survive. What is the correct way
This is the last in the series of exit cleanup patches for a while. I'm
tackling ffmpeg next, and expect that to take some time. This is the
biggest of the series as well, it removes premature exit()'s from the
player and aborts the playing instead with an informational dialog for
the user. It
Taylor Jacob wrote:
I have seen some prople having problems compiling the dvb-kernel drivers to work
with the Air2PC.. 2.6.10 came out on Dec-24-2004..
I just downloaded 2.6.10, got the dvb-kernel cvs code from today..
Ran the makelinks script in dvb-kernel to inject the driver into the linux
On Sunday 26 December 2004 10:07 pm, Mark Spieth wrote:
as per isaacs observation, continually asking the db for info is a very bad
thing. so here is a patch with an estimating version. works ok I think but
not perfect of course. also included is improved video smoothness in
playback esp when
On Sunday 26 December 2004 05:08 pm, vitold wrote:
Hi,
I'am a litle conffused, I realise that my first patch wasn't very clear,
there was a second fix for a segfault than can occur at the end of the
playback becose of a bad sycronization of two threads:
I think a better fix would be to
On Tuesday 28 December 2004 01:37 am, tommy wrote:
The pointer manipulation in the two lines of code are incorrect for k8
processors, they were causing the frontend to segfault on playback.
This hasn't been tested on a 32 bit processor, so if someone could make
sure it compiles without
On Tuesday 28 December 2004 09:13 am, Paul wrote:
Isaac wrote:
The attached patch adds a metadata editor to MythMusic which
allows you to save the changes optionally to both the database and
the file. You access the editor from the playback screen by pressing
the INFO button. Useful
On Tuesday 28 December 2004 11:19 am, Paul wrote:
This patch fixes the problem of MythWeather not always being able
to find the correct weather icons after recent changes by J. Donovan
to allow weather icons to be in a weather sub directory under the theme
directory, which only Titivillus has
On Tuesday 28 December 2004 11:51 pm, Jack Porter wrote:
Hi,
This is a trivial patch that adds missing fromUtf8 conversions for
channel name and callsign in a few places. Some places were already
doing the conversion but others were missing, causing junk in some menus
such as the Watch
When i was developing the windows filters i found that some files have video
timecodes thats realy wierd, they start realy high, then increase for a few
frames, and then reset back to zero and then continue normaly.
I had to develop a workaround for this because windows (actualy directshow)
the stuff in GetFrame is not for the end of buffer detect.
this was for smoothness in playback when videooutrate playrate
yes that bit is complicated.
the stuff in IsLiveAndNearEnd is quite simple I thought and works quite
well.
I can redo the patch without the stuff in GetFrame if you like but
I just noticed this error in my mythbackend.log. This is a fresh
build from cvs. Any ideas as to what this is?
I think your mythbackend binary is calling a different
libmyth than it was compiled with. Forgot to do a make install?
Different PREFIX to your runtime LD_LIB paths?
--
Nigel
i was lookin the code of some complex sources of
mythtv (such as main.cpp) and i wonder why is that
there are no comments of the code (or very few).
i'm tryin' to understand it (the code, i mean) and i
realy can't!
¿thus anybody have a diagram of the layers on myht?
it's a real pain in the
Doug Larrick [EMAIL PROTECTED] writes:
Your big issue here is that WGBHDT3 and WGBHDT4 don't really exist.
Unsubscribe from them at zap2it, and remove them. WHDHDT2 is another
phantom. The others look all right to me (I'm in the Boston area), but
you have duplicate channel numbers which
Steven wrote:
Thanks for that. I think you have to resend the patch as a unified diff
to get it committed here (diff -u).
Doh! Can you tell I'm new here?
Maybe it would be nice to also show the string value from the database
that is shown in the guide when there is nothing on? (=the result from
Sorry, here's the patch again in unified diff format.
Jack Porter wrote:
Hi,
When using Browse mode while watching live tv, if you pass a channel for
which there is no current program data, junk is displayed instead of the
channel name/number.
This patch makes
On Wednesday 29 December 2004 03:08 pm, Daniel Thor Kristjansson wrote:
This removes exit()'s in datadirect.cpp, and changes grabLineupsOnly(),
grabData() and grabAllData() so that they return true on successful
completion, and false on unsuccessful completion. It also adds checks
for error
This patch should be self-explanatory.
Cheers,
Kyle
Index: libs/libmythtv/channelbase.cpp
===
RCS file: /var/lib/mythcvs/mythtv/libs/libmythtv/channelbase.cpp,v
retrieving revision 1.10
diff -u -r1.10 channelbase.cpp
---
Le 28.12.2004 16:12:56, Daniel Thor Kristjansson a écrit :
On Tue, 28 Dec 2004, Isaac Richards wrote:
] This patch removes the -mfpmath=sse performance recommendation, but
] keeps the additional archs, such as the winchip, the updated C3
info,
] and other Intel and AMD processors.
]
]Have you
These are now in CVS, but without any MYTH_PROTO_VERSION change.
This should be safe if the frontend is newer than the backend,
since they are new queries that should be ignored.
--
Nigel Pearson, [EMAIL PROTECTED]| Let's see how Spike is going
Telstra BID, Sydney, Australia |
42 matches
Mail list logo