On Mon, 20 Apr 2009, Yang Tse wrote:

> All of the above make ne now think that curl_socklen_t should probably be
> defined in curlbuild.h and checked in curlrules.h to avoid nasty mismatches.
>
> Thoughts, besides 'what a pain'?

There are similar thoughts involving HPUX and a curiousity on how they ended
up in this mess... and how many possible libcurl users there are that use HPUX
etc.

But then I don't think putting curl_socklen_t in the curlbuild.h header for
mismatch check ability is a very big deal. If it can save a poor user from
getting into trouble, let's do it. I can't see that it will harm anyone.

-- 
  / daniel.haxx.se