Why do we not know who wrote DOS 3?

[walter_J64bit]

My Dad got my first 1050 used,DOS3 was pack in DOS. I did have a game menu that worked with DOS3 but Ive lost it, too bad it was nice game menu it had the Atari fuji logo at the top of the screen, under the fuji logo you seen the game select menu.

[Rybags]

Any chance of a transcript version of the interview... I just can't be bothered much with audio interviews.

 

I also recall seeing a nice menu with the Fuji, might even still have it and do remember it not being a Dos2 structure which along with being larger in size than most menus made it unattractive.

[_The Doctor__]

Dos 3 had an access dos 2 function... most people didn't bother with it thought I had a few disk autorun the utility... later DOS 2.5 had an access DOS3 utility.... soooo...

My first XF didn't come with DOS XE.... my second XF came with DOSXE... the menu systems between DOS3 and DOSXE had the same feel and look to them...

[+MrFish]

Listen to the Antic podcast interview with Bill. He confirms he wrote it.

 

Yep, I interviewed Bill, and he said that he wrote DOS 3.

 

He's an honest man to admit that.

[Savetz]

 

 

He's an honest man to admit that.

 

Statute of limitations has passed

[+MrFish]

Statute of limitations has passed

 

That's right, but now we all know where to fire the arrows.

Edited December 23, 2014 by MrFish

[bbking67]

I remember having a bunch of DOS 3 disks at some point and notching them to use for other things. I never actually came across anyone who used it. Most Atari users I knew used Atari 2.0S, 2.5 or some third party DOS (SmartDOS seemed especially popular).

[Bryan]

Bill doesn't seem very fond of DOS 1/2. It was a limiting design, but not having to buffer a sector table does save you some RAM. Since the Atari drives had a CPU in them, they could have offloaded some of the disk management with a few extra commands.

[flashjazzcat]

DOS 3's filesystem was much better than DOS 2's awful chain-links aside from the wasteful cluster size, but perhaps they were so large to enable the FAT to fit in a single sector (which I assume it did).

[+Ripdubski]

In the interview bill mentions the f/s could handle disks up to 16MB through plug in drivers that Atari never asked him to write.

[walter_J64bit]

In the interview bill mentions the f/s could handle disks up to 16MB through plug in drivers that Atari never asked him to write.

16MB disk under DOS 3 that is just insane!

[sup8pdct]

I have just heard the interview with bill wilkinson. The time lines he talks about in regards to DOS 3 don't make sense.
He said dos 3 was written near the end of oss and last gasp for the 8 bit line which was near 1988. DOS3 came out with the
1050 in 1983. I personally don't think bill wrote DOS3 and it would seam to me that he may be confusing DOS3 and DOSXE.

Maybe someone can email him to confirm my thoughts.

James

[evilmoo]

Perhaps he meant "the end" as in the end of their ownership by Warner Brothers.

[ricortes]

I had a different memory of the way things went down too. Something like DOS 3 was an in house replacement for DOS II. I believe Bill wrote an article or two in his Insight Atari column about how there were problems with its minimum allocation size of 1k. He explained something along the line of a 40 byte ML subroutine file would take 1024 bytes on disk and a 1025 byte file would take 2048 of disk space. I think he wrote a patch for DOS II or 3 and later that year in his annual 'things I wasted time on' wrote that DOS 2.5 came out so soon after the column there was no need for the whole exercise.

 

Likewise I seem to recall an article or post about the last days of OSS where Bill was laboring away on a 1200 XL computer writing the last bits of DOS XE. Not sure about the full history but I have it Shepardson Microsystems/OSS did BASIC, DOS I, DOS II, DOS 2.5, and DOS XE. Some of this is only based on time of release and not anything I can specifically cite like Compute magazine.

[thorfdbg]

DOS 3's filesystem was much better than DOS 2's awful chain-links aside from the wasteful cluster size, but perhaps they were so large to enable the FAT to fit in a single sector (which I assume it did).

Yes, they were - that's precisely the reason. Dos 2.0S had the convention using one buffer per drive, one buffer per open file. The drive buffer keeps the VTOC, the sector allocation table. The file buffer buffers the data read or written to disk. DOS 3 tried to continue with this convention, hence they tried to squeeze the VTOC into a single sector to cut the memory usage down to 128 bytes/sector. Unfortunately, Dos 3 keeps the file linkage in the VTOC, requiring 8 bits (one byte) per linkage, instead of the one bit per sector for Dos 2. Hence, the number of bits per sector went up by a factor of eight, and hence the number of blocks per sector had to go up by a factor of eight, too, to squeeze the entire FAT into a single sector.

 

One way or another, this was needlessly wasteful for the already small Atari disks, and later attempts for the Dos (Dos 4?) had again a one block per sector convention, though required then more than a single sector for the VTOC.

 

It is more or less a trade-off one had to make. However, the Dos 2 "chain linkage" was really a pretty bad design decision as it limited random access. Unfortunately, it was so popular it wasn't possible to change it anymore.

[fujidude]

Forward looking or no, two huge problems:

1. No backwards compatability.

2. Wasting space in a system where you really don't have a lot of space to start with.

 

Now then, if #1 was the only problem, people might have been willing to slowly migrate to it over time I think.