|
List Info
Thread: Re: Client handling for wrong Range field on Seekresponse
|
|
| Re: Client handling for wrong Range
field on Seekresponse |
  United States |
2008-03-12 15:12:15 |
The client has to use RTCP for sync for *any* stream for
which it does
not receive an RTP-Info with rtptime (live or otherwise). In
this case,
the server is sending the right times in RTP-Info rtptime
field, at
least at the initial start of playback - and it start time,
clearly
the client is using the RTP-Info time for sync, but that of
course
doesn't mean the client is not trying to use the SR time for
something
else, or using it for sync in the post-seek case. But I do
not know
offhand what all the client will use the SR rtp-time for or
what the
preference order would be between that and RTP-Info rtptime
at initial
startup and/or seek.
It is also possible that there is a problem with the
RTP-Info rtptime
and/or packet times after seek, or that there is something
that the
client is handling incorrectly with them.
Thanks,
Jamie
rajesh.rathinasamy nokia.com wrote:
> Jamie,
> You are right, the RTP timestamps in sender reports
are wrong. I
> thought client does RTCP time sync only for live
streams and not for
> on-demand clips. Let me check that.
>
> I did not get what you are inferring with wrong RTP
timestamps in sender
> reports. Is that the cause for current problem or is
that an another
> problem with this server ?
>
> Please let me know if you need the complete ipdump.
>
> Thanks,
> Rajesh.
>
>
>> -----Original Message-----
>> From: ext Jamie Gordon [mailto:jgordon real.com]
>> Sent: Wednesday, March 12, 2008 1:52 PM
>> To: Rathinasamy Rajesh (Nokia-D-MSW/Dallas)
>> Cc: ehyche real.com; nokia-private-dev helixcommunity.org;
>> protocol-dev helixcommunity.org
>> Subject: Re: [Protocol-dev] RE: [Nokia-private-dev]
Client
>> handling for wrong Range field on Seekresponse
>>
>> If you look at the RTP/RTCP traffic instead of just
the RTSP,
>> that server is severely screwing up the RTP time in
the sender reports.
>>
>> rajesh.rathinasamy nokia.com wrote:
>>> Hi Eric,
>>> That was my initial guess, but it does not
look like that.
>>> Whereever you seek to, server sends data from
previous play position
>>> when it was paused.
>>>
>>> I seeked back to zero while playing 25 sec and
the play response
>>> starts from 29. So it is not the key frame
issue. Seems more like
>>> server does not respect the play message with a
new range field and
>>> uses its time measurement to resume playback.
>>>
>>> Attaching the ip dump from microcore client.
>>>
>>> Thanks,
>>> Rajesh.
>>>
>>>
>>>> -----Original Message-----
>>>> From: ext Eric Hyche [mailto:ehyche real.com]
>>>> Sent: Wednesday, March 12, 2008 10:46 AM
>>>> To: Rathinasamy Rajesh
(Nokia-D-MSW/Dallas);
>>>> nokia-private-dev helixcommunity.org;
>> protocol-dev helixcommunity.org
>>>> Subject: RE: [Nokia-private-dev] Client
handling for wrong Range
>>>> field on Seekresponse
>>>>
>>>>
>>>> I believe what is happening is that when
you tell the
>> server to seek
>>>> to a certain point in the video (say 45
seconds), then the
>> fileformat
>>>> will look for the first keyframe prior to
45 seconds,
>> because it HAS
>>>> to start sending packets from there.
>>>>
>>>> I'm guessing that if you looked at the
file, you'd find that the
>>>> first keyframe prior to 45 seconds is at 16
seconds. If that's the
>>>> case, then the PLAY response is correct.
>>>>
>>>> Eric
>>>>
>>>>
=============================================
>>>> Eric Hyche (ehyche real.com)
>>>> Technical Lead
>>>> RealNetworks, Inc.
>>>>
>>>>> -----Original Message-----
>>>>> From: nokia-private-dev-bounces helixcommunity.org
>>>>> [mailto:nokia-private-dev-bounces helixcommunity.org] On Behalf Of
>>>>> rajesh.rathinasamy nokia.com
>>>>> Sent: Tuesday, March 11, 2008 12:35 PM
>>>>> To: nokia-private-dev helixcommunity.org;
>>>>> protocol-dev helixcommunity.org
>>>>> Subject: [Nokia-private-dev] Client
handling for wrong Range
>>>> field on
>>>>> Seekresponse
>>>>>
>>>>> Hi,
>>>>> When we try to seek from a
particular server
>>>>>
(rtsp://207.154.19.148:554/CL0010508556_6284d5ee0.3gp),
>> the servers
>>>>> play (Seek) response comes back with a
range field which
>> is not the
>>>>> one requested for in the play request.
The range field seems to be
>>>>> same as where the stream was paused
rather than the seeked
>>>> position.
>>>>> Currently client stays in reloading
until the data for the
>> (client)
>>>>> seeked position shows up and plays from
where it was seeked to.
>>>>>
>>>>> For example,
>>>>> * Stream paused at 15 sec
>>>>> * Seeked to 45 sec (Sent play)
>>>>> * Play response has wrong Range field (
Range: 16-100)
>>>>> * Helix client stays in reloading until
data for 45 secs reaches
>>>>> client and starts to play from there.
>>>>>
>>>>> Backward seek results in video being
frozen and no audio
>> for a while
>>>>> and later plays the received data.
>>>>>
>>>>> We would like to know what should be
the correct behaviour ?
>>>>> Should Helix client/ Player adjust the
start time and play
>>> >from where
>>>>> server directed it to play ignoring
user command or the current
>>>>> behaviour of staying in reloading for a
while is the correct
>>>> behaviour
>>>>> ?
>>>>>
>>>>> Branch info: 221CayS &
SymbianMMF_rel
>>>>>
>>>>> Tried Head build splay which also shows
the same behaviour
>>>> as 221CayS.
>>>>> Thanks,
>>>>> Rajesh.
>>>>>
>>>>>
>>>>
>>>>
>>
------------------------------------------------------------
---------
>>>> ---
>>>>
>>>>
_______________________________________________
>>>> Protocol-dev mailing list
>>>> Protocol-dev helixcommunity.org
>>>> http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
_______________________________________________
Protocol-dev mailing list
Protocol-dev helixcommunity.org
http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
|
|
| RE: Client handling for wrong Range
field on Seekresponse |

|
2008-03-12 16:23:58 |
Jamie / Eric,
RTP-Info and RTP packets seems to be fine after seek and
they match.
I'm trying to validate what should be client's behaviour in
case of
server sending different Range in response to play than what
was
requested. Should the client adjust its timeline to what
server dictated
or stick to what the user has dictated ?
For example:
User seeks to 50 sec after 10 sec of playback, but servers
reponds with
data from 12 sec with the range in response indicating that
data is
coming from 12 sec. Should client/player adjust the timeline
to 12 sec
and start playing from there or stay in loading phase untill
data for 50
sec is received and play from there.
Thanks,
Rajesh.
>-----Original Message-----
>From: ext Jamie Gordon [mailto:jgordon real.com]
>Sent: Wednesday, March 12, 2008 3:12 PM
>To: Rathinasamy Rajesh (Nokia-D-MSW/Dallas)
>Cc: ehyche real.com; nokia-private-dev helixcommunity.org;
>protocol-dev helixcommunity.org
>Subject: Re: [Protocol-dev] RE: [Nokia-private-dev]
Client
>handling for wrong Range field on Seekresponse
>
>The client has to use RTCP for sync for *any* stream for
which
>it does not receive an RTP-Info with rtptime (live or
>otherwise). In this case, the server is sending the
right
>times in RTP-Info rtptime field, at least at the initial
start
>of playback - and it start time, clearly the client is
using
>the RTP-Info time for sync, but that of course doesn't
mean
>the client is not trying to use the SR time for
something
>else, or using it for sync in the post-seek case. But I
do not
>know offhand what all the client will use the SR
rtp-time for
>or what the preference order would be between that and
>RTP-Info rtptime at initial startup and/or seek.
>
>It is also possible that there is a problem with the
RTP-Info
>rtptime and/or packet times after seek, or that there is
>something that the client is handling incorrectly with
them.
>
>Thanks,
>Jamie
>
>rajesh.rathinasamy nokia.com wrote:
>> Jamie,
>> You are right, the RTP timestamps in sender
reports are wrong. I
>> thought client does RTCP time sync only for live
streams and not for
>> on-demand clips. Let me check that.
>>
>> I did not get what you are inferring with wrong RTP
timestamps in
>> sender reports. Is that the cause for current
problem or is that an
>> another problem with this server ?
>>
>> Please let me know if you need the complete
ipdump.
>>
>> Thanks,
>> Rajesh.
>>
>>
>>> -----Original Message-----
>>> From: ext Jamie Gordon [mailto:jgordon real.com]
>>> Sent: Wednesday, March 12, 2008 1:52 PM
>>> To: Rathinasamy Rajesh (Nokia-D-MSW/Dallas)
>>> Cc: ehyche real.com; nokia-private-dev helixcommunity.org;
>>> protocol-dev helixcommunity.org
>>> Subject: Re: [Protocol-dev] RE:
[Nokia-private-dev] Client handling
>>> for wrong Range field on Seekresponse
>>>
>>> If you look at the RTP/RTCP traffic instead of
just the RTSP, that
>>> server is severely screwing up the RTP time in
the sender reports.
>>>
>>> rajesh.rathinasamy nokia.com wrote:
>>>> Hi Eric,
>>>> That was my initial guess, but it does
not look like that.
>>>> Whereever you seek to, server sends data
from previous
>play position
>>>> when it was paused.
>>>>
>>>> I seeked back to zero while playing 25 sec
and the play response
>>>> starts from 29. So it is not the key frame
issue. Seems more like
>>>> server does not respect the play message
with a new range
>field and
>>>> uses its time measurement to resume
playback.
>>>>
>>>> Attaching the ip dump from microcore
client.
>>>>
>>>> Thanks,
>>>> Rajesh.
>>>>
>>>>
>>>>> -----Original Message-----
>>>>> From: ext Eric Hyche [mailto:ehyche real.com]
>>>>> Sent: Wednesday, March 12, 2008 10:46
AM
>>>>> To: Rathinasamy Rajesh
(Nokia-D-MSW/Dallas);
>>>>> nokia-private-dev helixcommunity.org;
>>> protocol-dev helixcommunity.org
>>>>> Subject: RE: [Nokia-private-dev] Client
handling for wrong Range
>>>>> field on Seekresponse
>>>>>
>>>>>
>>>>> I believe what is happening is that
when you tell the
>>> server to seek
>>>>> to a certain point in the video (say 45
seconds), then the
>>> fileformat
>>>>> will look for the first keyframe prior
to 45 seconds,
>>> because it HAS
>>>>> to start sending packets from there.
>>>>>
>>>>> I'm guessing that if you looked at the
file, you'd find that the
>>>>> first keyframe prior to 45 seconds is
at 16 seconds. If
>that's the
>>>>> case, then the PLAY response is
correct.
>>>>>
>>>>> Eric
>>>>>
>>>>>
=============================================
>>>>> Eric Hyche (ehyche real.com)
>>>>> Technical Lead
>>>>> RealNetworks, Inc.
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: nokia-private-dev-bounces helixcommunity.org
>>>>>>
[mailto:nokia-private-dev-bounces helixcommunity.org] On
>Behalf Of
>>>>>> rajesh.rathinasamy nokia.com
>>>>>> Sent: Tuesday, March 11, 2008 12:35
PM
>>>>>> To: nokia-private-dev helixcommunity.org;
>>>>>> protocol-dev helixcommunity.org
>>>>>> Subject: [Nokia-private-dev] Client
handling for wrong Range
>>>>> field on
>>>>>> Seekresponse
>>>>>>
>>>>>> Hi,
>>>>>> When we try to seek from a
particular server
>>>>>>
(rtsp://207.154.19.148:554/CL0010508556_6284d5ee0.3gp),
>>> the servers
>>>>>> play (Seek) response comes back
with a range field which
>>> is not the
>>>>>> one requested for in the play
request. The range field
>seems to be
>>>>>> same as where the stream was paused
rather than the seeked
>>>>> position.
>>>>>> Currently client stays in reloading
until the data for the
>>> (client)
>>>>>> seeked position shows up and plays
from where it was seeked to.
>>>>>>
>>>>>> For example,
>>>>>> * Stream paused at 15 sec
>>>>>> * Seeked to 45 sec (Sent play)
>>>>>> * Play response has wrong Range
field ( Range: 16-100)
>>>>>> * Helix client stays in reloading
until data for 45 secs reaches
>>>>>> client and starts to play from
there.
>>>>>>
>>>>>> Backward seek results in video
being frozen and no audio
>>> for a while
>>>>>> and later plays the received data.
>>>>>>
>>>>>> We would like to know what should
be the correct behaviour ?
>>>>>> Should Helix client/ Player adjust
the start time and play
>>>> >from where
>>>>>> server directed it to play ignoring
user command or the current
>>>>>> behaviour of staying in reloading
for a while is the correct
>>>>> behaviour
>>>>>> ?
>>>>>>
>>>>>> Branch info: 221CayS &
SymbianMMF_rel
>>>>>>
>>>>>> Tried Head build splay which also
shows the same behaviour
>>>>> as 221CayS.
>>>>>> Thanks,
>>>>>> Rajesh.
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>
>--------------------------------------------------------
-------------
>>>>> ---
>>>>>
>>>>>
_______________________________________________
>>>>> Protocol-dev mailing list
>>>>> Protocol-dev helixcommunity.org
>>>>> http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
>
_______________________________________________
Protocol-dev mailing list
Protocol-dev helixcommunity.org
http://lists.helixcommunity.org/mailman/listinfo/pr
otocol-dev
|
|
[1-2]
|
|