On Thu, 21 Feb 2002, [iso-8859-15] Andrés García wrote:

> When using the following to resume a download.
>
> curl -C 5000 -O http://www.domain.com/something.tgz
>
> I had already donwloaded, for example 6000 bytes, and I was expecting curl
> to start downloading at the given offset overwriting the file from 5000 to
> 6000, just in case the first download was corrupted at the end.
>
> Instead it appends the new download at the end the file, so I end up with a
> corrupted file.
>
> Is that the expected behaviour?

That is the expected behaviour, yes. The continue-at support just asks to
continue the download from a given position, it doesn't really care about
what you already might have downloaded before or not.

What you're asking for here is a feature that curl doesn't have. I agree that
it might be useful and I know some other download tools supports this kind of
operation.

> I have tried to compile 7.9.5-pre2 in Linux but the configure fails:

Yeah, Kevin Roth also told me. I'll make another pre-release available really
soon, you might be able to "fix" this as I described here:

        http://curl.haxx.se/mail/archive-2002-02/0137.html

-- 
    Daniel Stenberg -- curl groks URLs -- http://curl.haxx.se/