rohc team mailing list archive

[Didier Barvaux <didier@xxxxxxxxxxx>]

Hello,

>    Yes, I think we should.  Unfortunately, we have made significant
> modifications in the library to accommodate our kernel modules with
> other compression protocols.

If your changes are of general interest, it would be great if you could
provide me some patches for anyone to get your improvements.


>  I think we are going to look at your
> fixes and incorporate them, instead of upgrading the entire library.
> On another note, your fix to allow multiple options list structs,
> with rasing the wlsb_width parameter significantly reduce our
> errors. 

Great to know! :)


> By the way, how high can wlsb_width be set to, and what is
> the tradeoff with a high wlsb_width value?

The W-LSB value must be a power of 2. There is no maximum except that
the rohc_comp_set_wlsb_window_width() function uses size_t for the
W-LSB width, so you cannot use a value that will overflow size_t on
your platform.

The advantage of a high W-LSB value is an increased robustness. The
drawback is a lower compression efficiency. The best trade-off depends
on your network quality and the traffic that you compress.

Regards,
Didier

Attachment: signature.asc
Description: PGP signature