Possible candidate on the Ethereal wish list:
Provide a general toggle where we have multi-line rendering of dissected
packets, one line per protocol. In this case, it would be easy to display
protocol-specific addresses (e.g., UDP ports) in the address column of the
relevant protocol. If the toggle is off, the output defaults to what we get
today. The output would then be similar to the packet dissection window, but
organized differently (per column).
I am aware that this new feature has a rather big impact.
Regards,
Olivier
> -----Original Message-----
> From: Guy Harris [mailto:guy@xxxxxxxxxxxx]
> Sent: donderdag 19 juni 2003 21:34
> To: Shaun Jackman
> Cc: ethereal-dev@xxxxxxxxxxxx
> Subject: Re: [Ethereal-dev] Address columns
>
>
>
> On Thursday, June 19, 2003, at 12:16PM, Shaun Jackman wrote:
>
> > Why is COL_INFO set using col_add_fstr, but address columns (like
> > COL_DEF_SRC) set using SET_ADDRESS(&pinfo->...)?
>
> Because, currently, the columns are set for all packets when the
> capture is read in, and that's expensive; setting the address is
> cheaper, and, by setting the address in the dissectors and
> then setting
> the column based on the address, it only sets the text once, rather
> than once per dissector.
>
> Note also that SET_ADDRESS isn't setting the column, it's setting the
> address; setting the address also makes it available to dissectors
> called later, and there are many dissectors that make use of that.
>
> > I have a higher level protocol (sits above UDP) that has its own
> > addressing scheme. I wanted to modify the address column to display
> > more
> > useful information. After trying col_add_fstr on COL_DEF_SRC and
> > meeting
> > with failure, I started looking through the source. It looks like it
> > can't be done without defining a new AT_* type. This is a simple
> > in-house plugin that currently builds without modifying the Ethereal
> > source. I'd like to keep it that way. Can this be done any
> other way?
>
> Set the address to an AT_NONE address (which means the column
> won't be
> set from the address), and then set the column with one of the
> "col_add" routines.
>
> Note, though, that you should do this *after* your dissector uses the
> addresses in the pinfo structure (or save the old value and
> use that),
> and that if you do this, any dissectors you call won't have access to
> the addresses.
>
> _______________________________________________
> Ethereal-dev mailing list
> Ethereal-dev@xxxxxxxxxxxx
> http://www.ethereal.com/mailman/listinfo/ethereal-dev
>
