Re: <ot>Setting up a hyper text link to a binary file</ot>

On Tue, 7 Apr 1998, Tim Jones wrote:

> IIRC Netscape had a section in their help system I found that stated 
> something like the above about  http: protocol for their browser.

I wouldn't have expected anything less from them... if the IT business 
ever dries up, their attitude will make them brilliant politicians or 
lawyers.

> True your server has to support the other transmission protocols ftp: 
> file:, gopher:  which may not be the case so...
 
It may not even go to the same server (program)... and the use of file:
(technically obsolete equivalent of ftp:) for non-local files is
*depreciated* these days.

> If you zip your spreadsheet and set the link to file using
> http://myserver/thing.zip transmission 
> protocol it's more likely to work as you want. 

That's good advice. The "problem" here, if you want to see it as a 
problem, is that HTTPd servers (IIS included), don't come with common 
Micro$oft file extensions premapped. (Zip isn't M$, so it usually is 
premapped.)

> Setting the MIME type in the browser doesn't always work as you
expect.
> For example I have  MIME type application/msword defined for all file
> .DOC and to launch MS word as an action.  But trying to get a file by
> http://my_server/~docs/example.doc from the local server doesn't
launch 
> MSword as expected but reads the binary file as html. Which is the
symptom
> I think people were  reporting for the spread sheet. 
> 
> I'm a Netscape user IE3/4 may behave differently.

That has nothing to do with it. This is a *very* common misconception.
The browser-side mappings are used ONLY if the server does not supply
a MIME type. But HTTP spec indicates that a server MUST supply a MIME
type -- if doesn't know explicitly, it is *required* to guess or use a 
default value.

What happens is this: The server gets a request for xyzzy.doc. It checks

its MIME mapping, doesn't find .doc and slaps on the default 
"text/plain". The browser gets back a file tagged as "text/plain" and 
that's the way it will be displayed. The browser *can't* make "educated"

guesses based on the original URL.

Why? Consider a CGI program, "xyzzy.exe". If the browser made an
educated 
based on the .exe extension, you'd never get the CGI program to return 
anything but application/octet-stream...

> >Short of that, you can also hope
> > that more people would learn to use their right mouse button...
> 
> To do what exactly?

To invoke "Save target as..." or equivalent (or however it is done in 
one's fave browser).

You see, it's like this: You have this nifty camping utensil called 
spoon-fork, because it has a spoon end and a fork end -- the idea being 
that the most suitable end is used for each cuisine.

Then you have loads of people trying to eat soup with the fork end, 
blaming the *cook* and asking: "Could make it a bit thicker?"

-- 
maxxon@swob.dna.fi (Mikko Kurki-Suonio) 	   | A pig who doesn't
fly
+358 50 5596411 GSM +358 9 80926 78/FAX 81/Voice   | is just an ordinary
pig.
Maininkitie 3C14 02320 ESPOO FINLAND | Hate me?    |	      - Porco
Rosso
http://www.swob.dna.fi/~maxxon/      | hateme.html |