Daniel,

We need to have it since we are shipping consumer applications and
they just need to work as best as possible. We do only need Mac and
Windows, so I don't know much on how various linux distros store this
kind of information.

Retrieving the system data on mac and windows is quite easy and only
a few pages of code. This can be just source code or a library. I
guess the library helps to solve link errors users who try and build
it most likely will have.

Determining if a .pac file exists and using this is a big job to do
properly. If we do this, we will most likely use the local
javascript engine on the users machine to parse this. So on Windows
we would use IE, on the Mac we would use WebKit, and for
compatibility and embedded systems we would implement the same thing
in SpikerMonkey. If we decide we can accept the 1mb of code for
spidermonkey in our application we may consider just using and
embedding this.

Has anyone considered implementing the autoproxy part before? I just
no know way but to use a full javascript engine to do this properly.

We will start on the code, and when we have something report back to
the list with a sample. I am not sure when we will do this, but
probably in the next month or two.

Matt Veenstra
tribalmedia

On Aug 28, 2007, at 4:12 AM, Daniel Stenberg wrote:

> On Mon, 27 Aug 2007, Matthew Veenstra wrote:
>
>> We will do our best to make this a chunk of code that can be added
>> in for those who need it. Should it be done in the same spirit as
>> the new curlx_getenv()? So curlx_getproxy(CURL *handle, char*
>> url, long authentication) // last one if needed.
>
> I'm interested to hear from more people on the desire of getting
> this actually inserted in libcurl.
>
> An alternative is to have a separate library with this function
> (which of course can be maintained, produced and provided under our
> umbrella still if we want to). I think such a proxy-finder function
> would be very useful for the (users of the) curl tool, and quite
> likely for numerous applications using libcurl.
>
> --
> Commercial curl and libcurl Technical Support: http://haxx.se/
> curl.html
Received on 2007-08-29