|Home » Resources » Rants
PLEASE READ THIS
Week of April 16, 2002
Hey Mr Microsoft,
Sorry for the spammy subject line, but I had to make sure this
letter got through and wasn't waylaid by one of your marketing
execs. I want you and you alone to see this letter, Mr Microsoft.
I've been doing some research since last I wrote. I just completed a
mail-order course in Visual Basic and I think I am about ready to
start writing some cool tight apps. Also, I took two weeks to learn
the Windows API. It's good to know the Windows API.
But anyway, I send these ideas to you, because your company is the
leading technological developer on the planet, and because you need
help innovating, after all that trouble you had in (the other)
Washington. Anyway, here they are.
One Computer One Root
I have been thinking about this a long time, and I think it's my
best idea. I mean, what do people in China do? Do they have drives
A: and C: and so forth, just like us? But they don't use our
alphabet, do they? They have their own alphabet, right? So how does
it feel to have our drive letters? Not good I bet.
There is an easy way out of this. I am surprised you guys haven't
thought of it by now, but if you act quickly you can still get
credit for the good idea.
What you do is you make a single root directory for every computer.
Then you make a sub-directory under root for example, that you call
'Floppy', and you 'mount' the floppy drive A: under this directory.
Your root directory is your C: drive, and if you have other drives,
then you 'mount' them under the root too.
And while we're at it - do we really need this horrid backslash? I
hate the backslash - it's so hard to get at on my computer. Can't we
use the forward slash instead?
When I run your COMP program, it often says 'Ten or more differences
- aborting' and that does me no good. I often want to know about the
ends of the files, if they're the same or not. So I was thinking,
couldn't you write a program called 'PMOC' (that's 'COMP' backwards)
which compares files from the end instead?
Nobody can do this yet, Mr Microsoft - you can be First in the
To Run or Not to Run?
Sometimes we have programs that can run but we don't want them to.
But we can double click on them by mistake and then we crash our
computers. That isn't good. So we have to rename the files to stop
them running when we click on them by accident. But that isn't good
either, as maybe we can't remember the original file name after a
OK. So why not have a file attribute 'X' which stands for 'eXecute'?
So only if this attribute is set can the file run?
And sometimes we want to exclude directories from searches, and if
we could use this 'X' bit on directories too, the 'X' would mean
'you can run this directory', that is, you can 'eXecute' this
directory, and if the 'X' bit isn't set, then you can't list it and
your directory is exempt from the search? So then we could use the
'X' on both program files and directories too.
And your file attributes right now allow you to read files but not
necessarily to write them, but never the other way around? Look at
it this way: Say we have this application, the users have to send
data to its files, but this data has to be hidden, so they can send
their data, but they're never allowed to see ALL the data as a
whole, because that would be INVASION OF PRIVACY we can say, so they
can only write, not read?
Can you do this too? Right now you have 'R' which stands for
'Read-Only' but the 'R' looks more like 'Read', and anyone can do
that, but you don't have a 'W' for 'Write', do you?
So use them both. And if you don't have a 'R' then you can't read a
file, and if you don't have a 'W' you can't write it.
As far as I know, there is no other company which has this idea
either, so once again Mr Microsoft - step on it!
Sometimes you need a program to run faster than others, and the
normal time slicing is not good enough. So if we could ourselves set
a file attribute bit so the program stays in memory more often...
Again, Mr Microsoft, I think if you act now, you can still be
When directories need to expand on disk, they fragment the disk free
space, so my suggestion here is that you use less information in the
Like today you have the file name, then the extension, then the file
attributes, then the three times, then the starting cluster - take
all that out.
Have a single file on every physical drive with all this information
in a big long list - call it the INDEX LIST FILE, or 'ifile' for
short. The ifile will be easy to read and maintain too - just think
Then instead of the directory information, all you have is the file
name and the INDEX into the ifile - that's it!
Oh - by the way - have you understood the implications here? Because
the file name is not directly tied to the physical file information
but to an INTERMEDIATE NUMBER (an index), ANY FILE WITH THE SAME
INDEX WILL IN EFFECT BE THE SAME FILE - even with a different name,
and even if the file is in a different directory!
This will of course not work on different physical drives, but still
There is a caveat too, you guys might have found it sooner or later,
I don't know, but when you delete a file, you don't actually delete
the file itself, you destroy the link to the ifile. Then when you
don't have any more links to the ifile, you delete the file and free
its index in the list.
This could be *very* cool, as you could traverse from one side of
the known universe to the other with a single 'chdir' command, just
like Picard does with his Beamer.
So the directories themselves would be a lot smaller, so they
wouldn't fragment disk free space as often, so the computers should
run faster on an average too.
Well that's my ideas. I hope you like them. You don't have to give
me money for them, I know you wouldn't give me any anyway, just give
me a little credit, ok? Maybe Microsoft Classic Board Games?