Jakub Zawadzki wrote:
On Tue, Sep 18, 2012 at 04:10:09PM -0400, Bill Meier wrote:
I haven't paid any attention to this change
("yet another ... passing data ...").
I tried to discuss this in [1].
I gather that it's not backwards compatible ?
Um, that would seem to mean that many private custom dissectors will
need to change (with a macro or whatever) when 1.10 is released (if not
before).
Idea was to replace all dissector_t variants into one, which definetely would break API.
Simpler version of patch I commited after almost week (night of 4/09 -> evening of 10/09).
It took me some time do to it, but I'm open to revert this changes and add new
*_dissector_t (instead of replacing), but do we really want 5 different dissector types:
old-style, new-style, data-style, heur, heur-data?
No worries... I'm all for progress and (AFAICS) this is progress (I
haven't liked the private_data stuff since I discovered, circa r34408,
that we had to restore it after an exception).
It does break things but that's just another problem to solve...