On Thu, 20 Apr 2006, Alexander Lazic wrote:

>> Lots of room for people to step forward and offer some patches! ;-)
>
> Yep i have started to look into the main.c and some other parts, what I have
> seen is that this changes are a mayor code && design change.

Well, yes but the tool is pretty simple as it is now so cramming in advanced
fetures in a simple loop like today is of course bound to require changes!

Given some more thoughts and planning, I would actually think that the curl
code would benefit from such a little reorg.

> curl url1 : --serialize -H $HEADER_FROM_PREVIOUS_CALL url2
> curl url1 : --serialize -d $FORM_FIELD:formvalue url2
> curl url1 : --serialize -d `get_form_field`:formvalue url2

I read you. These are mighty advanced features indeed. It'll take a good deal
of thinking to define how these would work and be used.

Personally, I don't see the benefit of trying to remove all external scripting
by adding every possible need to the tool itself. It just makes the curl tool
(and code) larger and more complex, and it will NEVER be able to completely
replace external scripting needs anyway.

But I'm certainly open for discussions and ideas.

I could very well imagine building a 'curlboo' (I picked a silly name to
reflect that the name itself doesn't matter) tool that would have all curl's
features and an additional bunch, just to keep curl somewhat simple and yet
offer complicated things in the new tool. They would of course share most of
the grunt code.

-- 
  Commercial curl and libcurl Technical Support: http://haxx.se/curl.html