[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [rohc] SigComp assembler and interpreter ?



Hi Carsten,
Thank you for your reply.

I understand the UDVM implementation is not the complete SigComp implementation, but as per my understanding it is the most significant part. The other components- compressors, state manager and the dispatchers (compression and decompression) can be built around the UDVM using open source components (e.g DEFLATE compression libraries etc.) Also, of these only the dispatchers would be higher layer/ application specific and thus may have IPR issues. Please correct me if I'm wrong.

So, I still have the following questions:
a) Has anyone provided a complete UDVM implementation that can be used in a SigComp module?
b) In your original implementation (of the -05 draft), how was the compressed message for "This is a test message.  What is the answer to life, the universe and everything ?" generated? Although the documentation says DEFLATE, I did not get the same result using the zlib DELFATE implementation. Did I miss something?
c) Is the decompression code for DEFLATE algorithm available in the UDVM instruction set or in bytecode?

Thanks once again.

Regards,
Simran



Carsten Bormann wrote:
Hi Simran,

2. Has there been any addition to the UDVM assembler/interpreter since March 2002? Specifically, has the implementation expanded to work with all the instructions in the UDVM instruction set? Also does anyone have any programming experience with the UDVM instruction set that can be shared?

Does "the UDVM assembler/interpreter" refer to my half-baked implementation of the -05 draft?
I haven't got around to work on this implementation to upgrade it with respect to the changes we made on the way to RFC3320 and the revised assembler syntax we agreed on for the implementers guide.

 3. I tried out the examples posted by Carsten and these worked perfectly. However, can anyone tell me how we got the compressed message in the first place?

A UDVM implementation is not a complete SigComp implementation.  In addition you need:
-- an interface to the higher layer protocol (e.g., SIP)
-- a state manager
-- a portfolio (>= 1) of compression algorithm implementations, and
-- UDVM code to send to the decompressor for each of these.

Clearly, these are significant pieces of work, some of which also have interesting IPR complications.

Gruesse, Carsten