[Live-devel] Patches in the Debian package

[Ross Finlayson]

>> 1/ The first 'patch' was apparently intended to remove some compiler
>> warnings.  It's harmless, but I recommend not applying it, because the
>> code that generates these compiler warnings might end up changing
>> sometime (which will break the patch).
> 
> Why do you cast integers/bytes into void pointers? Unless there is a
> reason, I think the patch should be applied?

Perhaps, but I'll probably just change the code to eliminate the compiler warnings.


>> 2/ The second 'patch' seems wrong to me.  I don't understand why any
>> system would define "struct ip_mreq_source", but not also define
>> "IP_ADD_SOURCE_MEMBERSHIP" (a constant that makes that structure
>> useful).  I recommend removing the patch.  If anyone feels that it is,
>> in fact, necessary, then they can post a message to this mailing list
>> (as they should have done in the first place!) explaining why they
>> think it's needed.
> 
> I found a reason in the logs: "ip_mreq_source is defined in all glibc
> not just on kfreebsd. Fix hurd FTBFS" I'll check if this patch is still
> needed.

See my next message...


>> 3/ The third 'patch' is definitely wrong, and should be removed.  The
>> whole point of deprecating the old, synchronous "RTSPClient" interface
>> is that any code that happens to depend upon the old interface needs
>> to explicitly be updated - by
>> "#define"ing RTSPCLIENT_SYNCHRONOUS_INTERFACE - so that it can
>> continue using it.  But anyway, no code that depends upon the old
>> interface is present anywhere in the LIVE555 distribution, or (I
>> presume) anywhere else in the Debian distribution either.
> 
> I see a problem here: We build a static/shared library without the
> deprecated old, synchronous RTSPClient interface. If a user defines
> RTSPCLIENT_SYNCHRONOUS_INTERFACE, he will get the old header functions,
> but linking with the static/shared library will fail.

There's no problem here.  In lots of places in the code, there are pieces of code that are #ifdef'd out by default.  (E.g., these often add extra diagnostic output to help with debugging.)  It's always been understood that if a developer wants to enable these pieces of code, then they need to build the library themselves - with appropriate compile-time flags - rather than rely upon a pre-built library.  This is no different.  So the third patch should be removed.

Ross Finlayson
Live Networks, Inc.
http://www.live555.com/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.live555.com/pipermail/live-devel/attachments/20121224/a569c86b/attachment-0001.html>