On Tue, 10 Jan 2006, Mike Jean wrote:

> I can use a different program for diff'ing if you like. It's no problem. You
> have to say what you use and the arguments. But I think my 'diff' will work
> too. I used the wrong arguments, I think. I made a 'unified' diff. A long
> time ago I used it for linux kernel patches.

I prefer unified diffs. I didn't exactly spot what in the format that made
your diff not comply, but my patch tool whined! ;-O

> (It's the diff program from GNU linux I use. It is only a win32 build.)

Weird. Well, let's ignore it and assume it was just me who messed up somehow.

>> I'm thinking that we should (for the non-HTTP doing CONNECT case), allocate
>> a fresh http struct to use for the actual CONNECT procedure and then go
>> back to the "ordinary" struct once the CONNECT has succeeded.
>
> Yes, that is also my second solution. But I spent only 4 hours of my time
> for compiling, including in my own app and analyzing the libcurl code.
> Within that time I found the problems and made the patches. So I didn't have
> the time for a better patch or learning enough about the libcurl code for a
> better workaround.

I think you showed us that you understood things and knew how to duck for the
problem!

>> You think you're up to make such a change for the FTP case?
>
> I think I can make it. But I don't know, if I will have enough time for a
> good solution with enough test cases. If you have two weeks time for waiting
> I will make the patch.

No rush here. This bug must've been present for *ages* and I can't see why we
can't wait another two weeks before it gets fixed....

Thanks for considering!

(Oh, and I corrected your name in the changelog and credits.)

-- 
  Commercial curl and libcurl Technical Support: http://haxx.se/curl.html