a) With persistent connection the number of times name resolution is performed
   (for my application) per download is very small.

b) (For my application) we know the name we are resolving is available on the
   local machine.

c) For our application we will be downloading very large files hence ...
   (sort of a http/ftp sort of thingy).

Thanks!

  ||From eeyem_at_u.washington.edu Wed May 30 01:41:18 2001

  ||On Tue, May 29, 2001 at 08:01:28PM +0200, Daniel Stenberg wrote:
  ||> On 29 May 2001, Alan wrote:
  ||> > b) How difficult would it be to hack it ot expose the FD to use it in an
  ||> > eventloop with many instances of libcurl ?
  ||>
  ||> I once planned to offer a "low level" interface that would do that (which was
  ||> dumped due to lack of demand). It would certainly not be a trivial nor a
  ||> quick fix to have that implemented. Still, of course it would be far from
  ||> impossible.
  ||>
  ||> The only really sane way to use many instances of libcurl today is with
  ||> threads.

  ||Also notice that name resolution will block anyway, so exposing the fd
  ||won't gain you true asynchronicity unless you're only retrieving
  ||exclusively by IP.

  ||--
  ||Evan Martin - martine_at_speakeasy.org
  || http://www.speakeasy.org/~martine

_______________________________________________
Curl-library mailing list
Curl-library_at_lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/curl-library
Received on 2001-05-30