I think I know why you were treated the way you were, and people
submitting patches can learn from your experience.
If you look back at your original patch, it consists of 293 lines
which break down, to a loose approximation, into roughly 50% for
elimination of obsolete code, 40% stylistic changes, and maybe 10%
actual technical content.
This isn't a crime in and of itself, but it makes it extremely
difficult to figure out what material changes you've made to the code.
If you'd cut the 90% "fluff", the purpose of your changes might have
been more clear. In any case, you owed at least a word or two of
technical explanation. And "the following improves the performance of
NE* clones" isn't an explanation.
It's natural to be excited about a patch you've spent a lot of time
working on, especially when the patch represents an important
technical fact you've discovered. However, no one is served by a
patch that requires explanation and doesn't come with any, no one can
help but resent having stylistic considerations imposed on them, and
no one who has to pore through the aforementioned stylistic mumbo
jumbo is going to be very receptive to obscure fiddling performed
without explanation.
Your expectations were unrealistic, too. You wanted Paul to run your
patch on a spare copy of "ne.c", but if it had worked, what would that
prove? The junk in "ne.c" represents years of trying to get around
hundreds of crappy clones. Paul doesn't have hundreds of crappy
clones to try your patch on, and since you didn't explain why you
thought your patch should work, why would he bother even if he did?
For the record, after you explained your reasoning in a subsequent
message, I decided your "discovery" rung true, and I'm running a
version of your patch on my cheapo NE2000 clone for kicks.
And it hasn't locked up y
;)
Kevin <buhr@stat.wisc.edu>