From: Joerg Reuter DL1BKE (zdxwh@kerailya.tunkki.fi)
Date: Sat Feb 05 2000 - 04:13:01 EET
> If this is so important for you I will do it the compatible way.
You still don't get it, do? I don't write these posts to boost
my ego or to humilate you. API changes have to be done in a most
sensitive way
> I think the time needed for this discussion can be used more sensible.
> I will leave this socket-interface stuff to somebody else.
No, go ahead and share your ideas with us. We'll see what we can
implement and how. We have discussed quite a lot in the past how
we can improve it and discarded many ideas because it would mean
a nightmare for application developers, distribution vendors or
just to maintain. I'm defending binary compatibility vigorously
because I do not want to see a mega big patch in the end that Linus
refuses to apply because it was bloated, breaking things and
conceptional wrong in some places. This has happened before to
the ISDN people (who have a similar approach to change things than
you) and the GGI folks. They literally had to plug their patches
like a goose to get at least the undisputably parts in. And Linus
wasn't very fond of Matthias' patch last time.
The most important thing about programming in such a big project is
communication. My impression is that you are looking at the whole
issue almost solely from the German Flexnet point of view. But Packet
Radio is _different_ in every part of the world. And these different
ways are reasonable for the respective situation.
As interesting and good your concepts may be -- we need to walk a
small path between flexibility and conservativism to keep it working
under most circumstances. This is not an academic concern -- I've
been bitten by these problems many times before.
Joerg Reuter http://poboxes.com/jreuter/
And I make my way to where the warm scent of soil fills the evening air.
Everything is waiting quietly out there.... (Anne Clark)
This archive was generated by hypermail 2b29 : Sat Feb 05 2000 - 20:11:09 EET