On Sun, 28 Apr 2013, Felix E. Klee wrote:

> In the description for [CURLOPT_TIMEOUT][1], it says:
>
>> Pass a long as parameter containing the maximum time in seconds that
>> you allow the libcurl transfer operation to take.
>
> What exactly is the "libcurl transfer operation"?

The entire thing you ask libcurl to do for a specific URL. From the initial
creation of a socket, name resolving, TCP connection, data transfer etc. Until
that particular transfer is done.

> Our program uses cURL multi (multi_socket) to download from various sources
> in parallel. Now, we run into issues where individual downloads clog the
> download pipe because they take too long. This is dispite the fact that
> currently we allow up to 1000 parallel downloads.

I don't understand. Why would single transfers "clog" a parallel pipeline?
What is a parallel pipeline?

> To cancel an individual download after some seconds, we have to implement
> our own timout logic, right?

No, not if you're using a recent libcurl. Then the standard timeout features
should work just fine.

-- 
  / daniel.haxx.se
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette:  http://curl.haxx.se/mail/etiquette.html