ImageDisk project is canceled

[Tony Duell]

> I always thought that if there really is demand for a utility then atleast a
> token amount of money should be charged for that utility. You will find that
> people who claim to need changes to the utility but are unwilling to pay a

I think you're comparing Apples and IBMs here :-). It is one thing to 
expect _you_ to make changes to your program (and that arguably should be 
charged for), it's another to want to make those changes themselves...

> few bucks for a new revision should just be ignored, they remind me of
> friends who keep taking up your personal time fixing their computers as long

Are you seriously suggesting you think that the Internet would be better 
if everything was charged for? No free downloads of old manuals like on 
bitsavers, for example (somebody spent time scanning those, converting 
them to pdf, etc). No free schematics on hpmuseum.net (I spent hundreds, 
no make that thousands. of hours drawing those out), etc.

I have written a few (OK, very few) utilities, and have never charged for 
any of them, and nor do I ever intend to.

> very little if any for the software that makes the machine usable.  You do
> not owe the world the source code for your hard work period. If people want
> to use some of your code they should pay for it or write their own. Sales

Next time somebody here asks for a sevice manual, or a schematic, or a 
pinout, or... I will not bother to look it up for them. I will simply 
point out that you can deduce the necessary information from the machine, 
from data sheets available on the net (and you'll have to go and find 
them) and a few hours with a mulitmeter. I won't bother pointing out 
useful repair tricks, it took me time to discrover them, why should I 
help you. 

> should be a yardstick to measure if continuing the development of the
> utility is worth the effort. Sooner or later the people who needed the
> utility will already have it, at that point you can quit developing and
> destroy the source code or release it in any manner you like. Why do I get
> the feeling that people who push really hard for open sourcing just want to
> use your knowledge and time for free in their projects?

So now you're claiming that all users of linux and *BSD are plagarists.

Yes, I will admit I've used ideas from open-source programs in my own 
programs. Yes, I will also admit that I've looked at published schematics 
and thought 'hey, that's a neat trick to save a few ns of delay', and 
used something similar in my own design. If the trick is significant, it 
gets an acknowledgement of the form 'Using a mux to decode the signals 
{foo...barr} was inspired by the DEC KT11-C schematic'

That, IMHO, is a far cry from ripping off somebody's work. Or do you want
to re-invent the wheel every time. 

Can any programmer or electronic designer honestly claim to have never 
read somebody else's code/schematics and taken ideas from it? Because I 
would doubt it.

To get back to disk imaging, I may be stupid, but I do wonder if it's 
actually possible to write such a program to run under linux or any other 
multi-tasking system. 

It is relatively easy to write a program to dump an image back to a 
physical disk. I've done similar things several times for specific disk 
formats (in particular TRS-80 M2/M4 and HP LIF). To do the reverse -- to 
imagge a disk _once the format is known_ is also easy.

The problem comes with discovering the format. I've read and re-read the 
8272 and 765 data sheets. From what I can see the only way to see what 
sectors are on the current track is to use the ReadID command. This 
returns the ID bytes from the next sector header to pass under the head. 
There is no way to automatically dump all the IDs from all the sectors 
from index pulse to the next index pulse (if there is, can somebody 
please enlighten me).

On a single-tasking OS, you could submit another ReadID command as soon 
as you get the reuslts  from the current one, and keep on going until you 
read the same ID again. That way you will hopefully have read all the 
headers on a track (hope you don't miss any ;-)). On a multi-tasking 
machine it seems to be hopeless. Your task could be switched out during 
this operation, you will then miss some sectors, you will not discover 
all the IDs.

If I am missing something, please enlighten me (point me to the 
appropriate section of the Intel or NEC datasheets, for example. I have 
both). If not, then why did nobody point this out <n> days ago, in which 
case this whole flamefest could have been seriously reduced.

-tony