On Mon, 10 Aug 2009, Patrick Monnerat wrote:

> This will not add any protocol overhead for servers that are not on OS/400,
> or that start in name format 1 mode. In addition, if some other "strange"
> servers/OS are discovered in the future, the infrastructure to deal with
> would already be in place.
>
> Daniel, would you accept such a patch ?

I think it sounds like a fine approach that will make libcurl able to work
agsinst OS/400 servers and it might even function as a foundation to support
other esoteric platforms later on. All that while in the majority of all case
it won't add extra round-trips. I figure it will though, on systems that
report their paths without an initial slash, like windows.

So yes, I'll accept one. I think it does deserve a healthy amount of comments
though to explain the need and reasoning for this procedure.

-- 
  / daniel.haxx.se