Re: Dia-list digest, Vol 1 #758 - 5 msgs

[dialist <pvspam-dialist hacklab net>]

> What we're really talking about, I think, is what two names to use for
> these three things:
> 
> 1.  a zipped dia file
> 2.  an unzipped dia file
> 3.  a dia file

Unless of course you've split the world into only the first two cases.
It's a matter of how you personally have set up the taxonomy of file
types. If you have broken the universe of Dia files into type 1 and 2,
then ".gzdia" and ".dia" suffice.

> Granted, if I mailed you my (zipped) .dia file and you tried to open it
> with "view as text" relying your mime setup, you'd be disappointed.  But
> we took a few turns to get there: you're relying on my naming convention,
> and forcibly interposing yourself between the file and its natural
> editor/viewer, Dia.  

Sometimes I really do want to edit the .dia file in a text editor.
Really. You might not believe it, but use it to make a database schema
with 30 classes, and you'll start to want to edit the .dia file in a
text editor, too. Just one simple case: I want to change every blue
class to a green one. Or, I want these 8 classes to all be the exact
same shade of purple.

When you are faced with eight .dia files, and you wonder which are
gzipped (and thus need preprocessing before editing) it would be far
less tedious if there were *SOME* indication of which were zipped and
which weren't.

I guess .dia.gz vs. .gzdia doesn't matter to me much except on a
philosophical level where I don't want *ANY* extension at all, much less
two of them tacked together.

> > It's ugly. I'd actually say given the current mess we're in to create a
> > new extension ".gzdia" or whatever, and fix it when filesystems actually
> > store MIME types with files. (We'd do away with extensions entirely at
> > that time, would we not?)
> 
> I believe that's what Macs call a "resource fork" and it's been around 20
> years.  Obviously, that's not long enough to gain adoption in these
> parts.... ;-)

I was thinking how Mac did indeed make a first stab at this quite some
years ago. I remember their solution wasn't without problems, but it was
at least going down the right path.

Does OS/X still have the concept of holding the file TYPE separately
from the file NAME? And what happens if the metadata for file types gets
erased or otherwise corrupted? Is the file no longer usable until you
get another copy of the file (like it was back in the bad old days)?

And why do I reckon BeOS had this down pat?

-- 
Tim Ellis
Senior Database Architect
author, tedia2sql (tedia2sql.tigris.org)