RE: [rohc] Updating the Formal Notation draft

["Price, Richard" <richard.price@roke.co.uk>]

> I'm still not sure "Variable" is the right word for this at
> the notation level, and as you say we should investigate if
> we can avoid having this at all. The examples you have shown
> seem to be about compressing fields together, i.e. using
> another field as a reference for compressing a field. Would
> "Reference" be a potential word if we have to keep this?

Hi,

Referencing fields is certainly the main use of "temporary
variables", but there are some others that might be worth
taking into account.

In general, we use the variables to store any miscellaneous
information that an encoding method needs to compress a field,
but which isn't available in the uncompressed packet or in
the context.  I've had a go at listing all of the places where
temporary variables come in handy below:

1.  Using another field as a reference to compress a field

This is the example mentioned above.  Currently we store the
actual value of the field in a temporary variable, but it
would be ok to instead store a pointer to where the field can
be found in the uncompressed packet.  So the term "reference"
would be fine in this case.

2.  Using a "control" field to compress a field

In some cases the field that we want to use might be a
control field, i.e. one created by the compressor and not
found in the uncompressed packet.  An example for TCP/IP
compression would be the "Master Sequence Number".  I'm not
sure how we could do this as a reference, since the control
fields aren't stored anywhere in the uncompressed packet.

3.  Other values that might be needed for compression.

We also use temporary variables to store other values, which
can't really be considered to be "fields" at all (since they're
not part of the compressed or uncompressed packets).  Here are
some examples:

   a.  Remaining_Data : The number of bits remaining in the
       uncompressed packet.

   b.  Index : A pointer to the current context entry.

   c.  No_Update : Set to 1 if the encoding method is not
       supposed to update the context.

I'm not sure that "references" captures all of the above, but
there are probably better terms than "variables" which we could
use.  Comments welcome!

Regards,

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