Re: Fixing CURLOPT_RESOLVE
Date: Fri, 20 Mar 2015 02:07:11 -0400
On 3/14/2015 1:37 PM, Daniel Stenberg wrote:
> On Sun, 8 Mar 2015, Ray Satiro via curl-library wrote:
>>> I've started some work on fixing CURLOPT_RESOLVE. I have fixed it so
>>> that multiple addresses can be stored for each host:port. You can
>>> see a branch comparison at .
>> I didn't get any feedback on this so I just went ahead and wrote it
>> in the way I think is most correct.
> Thanks Ray, sorry for being slow to respond to this. Looks great!
> Can you figure out some basic unit tests we can add for this so that
> we can be reasonable sure the functions all work as intended after
> these changes?
> Also, you mention support for multiple addresses but I don't see any
> change in the docs regarding that and I think it should get some
> mention (as then I won't have to ask you how they are supposed to
I've never written a test for libcurl before but I'll take a shot at it.
Besides the doc issue I also have to look into properly handling this in
multi. Right now the hostpairs in each easy handle are loaded before
that handle's transfer but I think it would make more sense to load them
into the cache when they are added to the multi handle. What do you
think? This weekend I will have some time to look at these things, but I
doubt I will finish by then.
On 3/17/2015 7:05 AM, Stefan Bühler wrote:
> you cannot change existing DNS entries after they have been added, because the entries are possibly shared between multiple threads, and only the access to the cache itself (and the ->inuse counter) is synchronized.
> Best regards,
> Stefan Bühler
Thanks I am aware that inuse entries should not be modified and I had
accommodated for that in my implementation. See the comment block at
. What I wrote in that first e-mail was my initial impressions. You
can see all the changes so far at .
List admin: http://cool.haxx.se/list/listinfo/curl-library
Received on 2015-03-20