On Sun, 6 Jan 2008, Daniel Egger wrote:

>> But test case 1020 fails for me, and if I understood you right your test280
>> worked for you so I though that perhaps you can have a look at this?
>
> Hm, the (expected) test result in the case is incomplete. Range -9 means
> from index 0 to (including) index 9 which are 10 bytes, so the numbers from
> 1 to 0 in this testcase.
>
> I've no idea where the trailing 0 got lost and apologize for the omission.

Sorry, but that's not the problem. If I append a trailing 0 it still fails,
and it fails earlier than the data comparison. Reading log/stderr1020 after
the failure shows this:

$ cat log/stderr1020
22:51:49.338947 * RANGE the last 9 bytes
22:51:49.391581 * range-download from -9 to -1, totally 9 bytes
** Resuming transfer from byte position -9
   % Total % Received % Xferd Average Speed Time Time Time
Current
                                  Dload Upload Total Spent Left Speed
   0 9 0 0 0 0 0 0 --:--:-- --:--:-- --:--:--
022:51:49.426396 * Closing connection #0
22:51:49.436059 * Couldn't resume download

curl: (36) Couldn't resume download

So if you can run my patch with this test case it is really weird...

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