I think the current way it works is fine - assuming curl tries to infer the target path based on the path specified for the local file. That way, if you're uploading a bunch of files in differing directories, you can keep them separate.

For Golestan, I suggest changing the directory. SETLOCAL and ENDLOCAL in your batch scripts (if you use those) will come in handy, as you will be automatically placed back in the directory you started in if you try the following:

C:\TEMP> SETLOCAL
C:\TEMP> CD \Windows
C:\Windows> CURL.EXE ...
C:\Windows> ENDLOCAL
C:\TEMP>

--Kevin

>>> Daniel Stenberg <daniel_at_haxx.se> 1/09/01 9:07 AM >>>
On Fri, 5 Jan 2001, Golestan Radwan wrote:

> One thing, though. I passed the full path of the file to upload (e.g.:
> c:\try) and when I looked on the server the file was called "c:\try".
> It's not a problem for me, as I could easily change the directory and
> upload from the current one, but I thought this might be an issue for you
> to look into in case you think it's worth it.

Does anyone else have any opinion in this matter?

-- 
  Daniel Stenberg -- curl project maintainer -- http://curl.haxx.se/