On Wed, 21 Jul 1999, Luke Mewburn wrote:

> This is a tricky issue; when you select a page in a browser that gets
> redirected and you save that page, the saved name is usually the name
> of the new page.
> 
> I'm not sure how other command-line http retrieval tools work determine
> which name to save a redirected file as; do you have any experience
> here?

Agreed, it is a tricky issue, and no, I have no experience of any
other command line tools.

> In any case, using `ftp -o somefile URL' should always save the output
> to somefile, even if URL is redirected. (At least, my testing with
> the -current ftp reveals this). If the pkgsrc stuff uses -o and then
> checks the generated file, then I can't see what the problem is.

Try this: 

ftp http://gd.tuwien.ac.at/editors/davis/slang/v1.3/slang-1.3.7.tar.gz
-o slang-1.3.7.tar.gz

What it does on 1.4.1 is make a file "gd.html" in the current
directory.

The only problem I see with using -o in the pkgsrc stuff (except that
it didn't work!) is that pkgsrc has an option to use other command
line tools besides ftp, like "fetch". Who has "fetch"? It would be
interesting to see how fetch handles the above url, or whether it has
an -o option.

It might be nice if ftp had, say, an "-O" option, to mean: "always
save the output in the requested filename, regardless of redirection"
(or vice-versa, depending on what's the preferred default). Then you
could set FETCH_CMD='/usr/bin/ftp -O', and be done with it.