On Tue, Apr 08, 2003 at 09:14:00PM -0700, Jaime Fournier wrote:
> Yeah, I commented out the col_clear at the top
> dissector for afs.
That shouldn't make a difference.
> I have attached a diff of
> packet-afs.c of what I did to try to get it to work.
> There is also a trace of a create-file for afs.
> The problem is that despite having included the pinfo
> to each function, no additional col_appends seem to
> work.
Are you certain that col_append_str is even being *called*? I doubt it
is; the subdissectors for AFS are called only if "tree" is non-null,
and, on the first pass through the capture, when the columns are being
set, it's probably null (it'd only be non-null if a read filter, a color
filter, or a tap is in effect - if a protocol tree isn't going to be
used, there's no point in constructing it, and reading the capture will
go faster if you don't build it).
The dissector would need to be changed to call the sub-dissectors for
the operations regardless of whether "tree" is null; the sub-dissectors
will probably work correctly, albeit not as efficiently, if not changed,
as they'll still be doing the work to build the protocol tree but the
"proto_tree_add" routines will just return because they'll be passed a
null pointer. It might be more efficient to skip whatever processing
isn't needed in order to add to the Info column.