On Thu, Jun 14, 2001 at 10:03:02AM +0200, Andi Kleen wrote:
> Has there been any consensus reached yet on merging Pavel Mores' TCP
> graphing patch into ethereal CVS? (http://kostra.uh.cz/ethereal/) I'm
> currently using it with 0.8.16 and I'm happy with it; but would like
> to update to a newer versions preferable without having to port the
> patch all the time.
Hi Andi,
unfortunatelly no consensus has been reached yet. It's partly my fault
because I failed to submit the last batch of changes required by Guy
Harris. Actually, the changes are all (or almost all - I'll have to
review it once more) in my private CVS but I never published them. I
know this was a stupid thing to do but my reasons included:
- I don't have much spare time these days - obviously - who has? ;)
- I have to admit that I'm a bit lazy
- more importantly: the way the patch was accepted was a bit confusing
to me. When I submitted the patch I expected roughly one of the
following reactions:
1) ok, we'll include it in the main tree, just tighten some loose ends
2) we don't think this kind of functionality is appropriate for
ethereal and we'll remove it from the wishlist
3) we do want this functionality but your code is messed up too badly
to be included.
In case of 1) I would simply do what's needed. I case of 2) the code
would remain for my private use which is OK, at least I wouldn't have
to do changes that I personally don't care about. In case of 3) I
would start rewriting and cleaning the code until it could be
accepted. But it was none of the above. Guy just asked me to make
some (mostly compatibility) changes but the project maintainers (who
exactly are they, anyway?) failed to give me a clear statement about
what the future of the patch should be.
- it seems that very little people actually use it. Nobody is interested
in old tired TCP these days, I guess. ;) There's been a couple of
downloads and that's it. Actually, your mail is probably the first
positive sign that someone uses it. (Maybe an announcement on
ethereal-users would be appropriate?)
So it seemed to me it was a waste of precious time to try to force the
code that nobody uses to maintainers who aren't enthusiastic (at the
very least), either.
OK, enough whinning. ;-) I'm still willing to put effort into having the
code integrated with the official tree. However, I would appreciate
having a clear understanding of what the maintainers' position
concerning the TCP graphing code is and generally, what it's gonna take.
pvl