>>>I suggest the following change to event_cb in asiohiper.cpp. This is
somewhat similar to what we do in our client.

Brian:

That's very similar to the fix I've been playing with. I'm not totally
happy with it, yet, because A) I think the problem could happen with
writes, as well as reads (hough I'm sure reads would be the more typical
case); B) although g->still_running happens to track a single transfer in
the sample, and would thus manifest the right state in this limited case,
any serious code that uses asio and libcurl's multi interface is likely to
have many pending transfers; patching the sample in this way sets up those
who study it for failure or at least head-scratching.

Neither of these issues is a deal-breaker, however; certainly your patch
would address the immediate need.

I kind of want to do a more ambitious patch, though; I'd like to fix a
number of other messes in that sample. Better function names and comments
would make it a lot more helpful. But I don't know when I will have time to
do so.

--Daniel

-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html
Received on 2015-04-24