We should create static functions to get these values with stream header 
passed in to prevent duplication of code.  

On Wed, 1 Dec 2004, Jamie Gordon wrote:

> We are obviously referring to two different things as "MDP". :)
> We should be able to safely add a new stream header using this
> logic to CHeaderHandler::StreamHeaderReady and just look for that
> single header in both streamgroupmanager and packetstream if the
> PacketStream won't be needed for PPM delivery. If we need this
> value for PPM delivery it will have to have the same logic or
> call a common function somewhere in PPM code path.
> 
> Jamie
> 
> Vijay Shetty wrote:
> 
> > I think this requires a change in MDP too. MDP currently uses the 
> > "PreRoll" field from Stream header. We have a few options to resolve this,
> > 
> > - add the same logic in MDP (not good for maintenance, any change in the 
> > logic will require changing code at multiple places)
> > - extend the InputSource interface to provide this calculated value. ( 
> > this would require a little extra work in MDP to get to the right value)
> > - add an extra field to the stream header which has this value (this 
> > will require a trivial change in MDP)
> > 
> > thanks
> > 
> > Vijay
> > 
> > 
> > 
> > At 11:22 AM 12/1/2004, Jamie Gordon wrote:
> > 
> >> I am not able to fully test that logic, but it changes the preroll
> >> used to calculate the upshift threshhold to use the ActualPreroll.
> >>
> >> The other affected uses of this preroll value are
> >> X-InitPreDecBufPeriod calculation (for advertising to client) and
> >> Predata calculation. The predata is used only for checking
> >> predecbufsize in the stream selector, which should be updated to
> >> instead use x-predecbufsize when available.
> >>
> >> Jamie
> >>
> >> Go Hori wrote:
> >>
> >>> Jamie,
> >>> Can you speak to the implication of this change to our rate manager 
> >>> logic
> >>> when playing RM contents to microcore client?
> >>> On Mon, 29 Nov 2004, Jamie Gordon wrote:
> >>>
> >>>> Synopsis
> >>>> ========
> >>>> MDP preroll fixup.
> >>>>
> >>>> Branches: SERVER_CURRENT (HEAD), SERVER_11_0_STABLE
> >>>> Suggested Reviewer: ghori
> >>>>
> >>>>
> >>>> Description
> >>>> ===========
> >>>> Updates MDP preroll calculation to use, in order of preference as
> >>>> defined in the client side spec
> >>>> https://client.dev.prognet.com/wiki.cgi/PrerollHandling
> >>>>
> >>>> 1. ActualPreroll
> >>>> 2. X-InitPreDecBufPeriod + X-InitPostDecBufPeriod
> >>>> 3. Preroll
> >>>> 4. Default (1 sec)
> >>>>
> >>>>
> >>>> Files Affected
> >>>> ==============
> >>>> server/engine/inputsource/streamgroupmgr.cpp
> >>>>
> >>>>
> >>>> Testing Performed
> >>>> =================
> >>>> Unit Tests:
> >>>> - Verified (in debugger and X-InitPreDecBufPeriod in SDP)
> >>>> that the appropriate values are chosen depending on which values
> >>>> are present.
> >>>>
> >>>> Integration Tests:
> >>>> None
> >>>>
> >>>> Leak Tests:
> >>>> None
> >>>>
> >>>> Performance Tests:
> >>>> None
> >>>>
> >>>> Platforms Tested: win32-i386-vc7
> >>>> Build verified: linux-2.4-glibc6-i686, win32-i386-vc7,
> >>>> sunos-5.8-sparc-server
> >>>>
> >>>>
> >>>> QA Hints
> >>>> ===============
> >>>> - Can be tested by checking the calculated a=X-InitPreDecBufPeriod in
> >>>> the SDP.
> >>>>
> > 
> 

-- 
Go Hori
ghori at real.com