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

Re: [rohc] SigComp User Guide, deflate bytecode

Well, actually, I would think that these extra bytes are actually record marking bytes, but not SigComp record marking: they are the bytes generated by the ZLIB library when you use it for partial compression, i.e.: call its deflate method with the flush argument set to Z_SYNC_FLUSH to compress a partial block.

This lets your compressor compress successive messages while keeping the compression history of the previous messages, hence making the compression of further messages more efficient. This mode of operation is needed to implement dynamic compression as per RFC 3321 when using ZLIB to compress the data.

>From the SigComp standpoint, these ZLIB record marking bytes can then safely be suppressed from the compressed data as the decompression bytecode does not need them...

Herve.
Hi,
 
I hope I'm not too late but I just saw a posting dated from 14 Mar 2005 (see below). After reading the User Guide, I thought the extra "ff ff" may have been placed there for the sake of record marking in a stream-based application. Does this make sense ?
 
Jean-Pierre
 
 
 
Hi Herve,

>
> While I'm at it, just one more little question on that topic:
> what are the extra "0000 00ff ff00" bytes added at the end of
> the sample message provided in the draft? They do not seem to
> relate to actual compressed data, and the sample message
> decompresses just as well without them...

You're right, it does. 
They don't break anything by being there because although the sentinel of 7
bits of zero is not necessary (as the draft says) it just happens that in
this message the total number_of_bits_needed mod 8 = 1 so we do have the 7
bit sentinel to make the message byte aligned.  Any bytes beyond it are
ignored.

I'm not sure why the bytes are there and they're not needed, so I'll remove
them. 

Thanks for pointing it out.

Best regards,

Abbie
>
> Herve Siegrist.
>
>
> _______________________________________________
> Rohc mailing list
> Rohc at ietf.org
> https://www1.ietf.org/mailman/listinfo/rohc
>

_______________________________________________
Rohc mailing list
Rohc at ietf.org
https://www1.ietf.org/mailman/listinfo/rohc

_______________________________________________
Rohc mailing list
Rohc at ietf.org
https://www1.ietf.org/mailman/listinfo/rohc
  
_______________________________________________
Rohc mailing list
Rohc at ietf.org
https://www1.ietf.org/mailman/listinfo/rohc