Darren New wrote:
> Orchid XP v8 wrote:
>> Well that's not really a "file". That's a magic filename.
> 
> Wellll... You can argue about it. I assume you know how it worked 
> inside. But yeah, I get what you're saying. There wasn't a directory 
> entry for the device drivers as such. Which, when you think about it, 
> makes sense.
> 
> You could iterate over device drivers. They just weren't on the disk, so 
> it used different APIs.
> 
>> Instead, there are lots of roots like "FOO:", "BAR:", etc. This is 
>> more like MS-DOS pathname syntax.
> 
> I'll grant you that one.
> 
>> Now, the way it *works* is of course more like Unix...
> 
> No, the way it works isn't like either. :-) It's closer to a microkernel 
> than anything.
> 
The way it worked is that any program could send messages to any other 
program. If a program answered to a specific set of messages it would be 
recognized by the OS as a device. That meant that any user could write 
his own device and indeed I have seen quite a lot useful and slightly 
silly ones. It was a very powerful concept, e,g, if it was still alive 
today we could easily implement a POV: device such that copying a file 
into that device would render that file. Filepath syntax could be 
exploited to set various options. "copy benchmark.pov 
POV:/640/480/AA0.3/png" That is of course not more readable than current 
syntax, but you could also do that with drag and drop. Hmm, okay, not so 
much of an earth shattering improvement either.

Post a reply to this message