I haven't had a chance to do this yet. I needed to get back to the
customer as quickly as possible with a fix/work-around, and I didn't
want to have to put in a whole new libcurl version. So what I did was,
rather than setting CURLOPT_PROXY to NULL to disable the proxy, set it
to "", which I guess should have the same effect but is possibly a bit
safer. I sent the fix to the customer and as far as I know this has
solved the problem.
In the meantime, yeah I guess we should upgrade to 7.14, and hopefully
at some stage I'll get a chance to test this a bit more.
Thx,
James

>
> Daniel Stenberg wrote:
>
>> On Tue, 16 Aug 2005, James Bailey wrote:
>>
>>> Oh yeah, I forgot to mention: after the first fetch returns
>>> CURLE_COULDNT_RESOLVE_PROXY, the application immediately retries the
>>> fetch, but by-passing the proxy (i.e., sets CURLOPT_PROXY to NULL,
>>> then calls curl_easy_perform() once more). The crash then seems to
>>> happen on the next fetch.
>>
>>
>>
>> Can you please please pretty please with sugar on top test this exact
>> same procedure with 7.14.0 to see if this problem still exists or not?
>>
>

-- 
James Bailey
Team Lead - Voice Browser
Voxpilot Ltd,
6-9 Trinity St.,
Dublin 2
Office: +353 1 2091914
email: james.bailey_at_voxpilot.com <mailto:james.bailey_at_voxpilot.com>