On Fri, 13 May 2011, David Stuart wrote:
> Here's a 7zip file containing the changed files for my patch
Please send us (unified) patches for changes as made with diff -u or plain
git. Complete replacement files are inconvenient. I've attached the patch to
this mail, generated from your files compared to what's in git right now.
Then we will also ask you to provide separate patches for separate changes so
that we can see exactly what you need to change to fix this specific problem.
There might be other things to fix and change as well, but they must be
provided in separate patches.
> using the 1.7.4 file on the website as the base.
Did you try using a daily snapshot of something more recent?
> Maybe someone could have a look at it?
I know nothing about the Windows specifics so I can't comment on those. I
assume regular old IPv4 and IPv6+IPv4 setups still work fine even with your
improvements?
The fact that you've modified an older version does make the patch flawed
though, as it removes things we've added during the last few months. That's
wrong. (My fault though for using the wrong version to generate the diff
from.)
Can you explain the reasoning by the change of try_config() ? To me it looks
like you've simply removed existing logic. It also seems irrelevant to your
windows improvements.
Your edits of advance_tcp_send_queue() is very hard to read since you
re-indented the function making just about the entire function modified but
most of the logic is ther same... The reindent seems irrelevant.
I assume you changed the struct field named 'try' to 'try_count' for some
weird compiler thinking it was a C++ compiler? Was this really a relevant
change?
> I'm also wondering whether I need to filter out some of the link local
> multicast/broadcast IPv6 addresses that seem to be returned in the DNS
> server list for Windows.
Hm, what will be the effect if you don't?
-- / daniel.haxx.se