24 Sep
2019
24 Sep
'19
12:18 p.m.
How did MS-DOS decide that two seconds was the amount of time to keep
the floppy disk cache valid?
https://devblogs.microsoft.com/oldnewthing/20190924-00/?p=102915
24 Sep
24 Sep
12:40 p.m.
I meant to say Raymond Chen and it didn't notice the spell check change
Sometimes modern technology sucks.
On Tue, Sep 24, 2019 at 3:18 PM Christian Liendo <cliendo at gmail.com> wrote:
How did MS-DOS decide that two seconds was the amount of time to keep
the floppy disk cache valid?
https://devblogs.microsoft.com/oldnewthing/20190924-00/?p=102915
12:51 p.m.
On 9/24/19 12:40 PM, Christian Liendo via cctalk wrote:
I meant to say Raymond Chen and it didn't notice
the spell check change
Sometimes modern technology sucks.
On Tue, Sep 24, 2019 at 3:18 PM Christian Liendo <cliendo at gmail.com> wrote:
>
> How did MS-DOS decide that two seconds was the amount of time to keep
> the floppy disk cache valid?
>
> https://devblogs.microsoft.com/oldnewthing/20190924-00/?p=102915
I don't quite get the bit about the NAND gate, but back in the days of
5.25" floppies there was a way around that one.
One need only monitor the drive write-protect status every tick or so.
If it toggles, invalidate the cache.
I implemented it on a 8085 system, checking the write protect status
every 250 msec. The drive needn't be spinning. If you insert or remove
a 5.25" disk, the status will change.
We even went a bit further and checked to see if the disk contained any
files open for writing. If so, we sounded an alert and froze the system
until the subject disk was re-inserted. This reduced the number of
disk-swapping related errors essentially to zero.
--Chuck
4:13 p.m.
On Tue, 24 Sep 2019, Chuck Guzis via cctalk wrote:
I don't quite get the bit about the NAND gate, but
back in the days of
5.25" floppies there was a way around that one.
One need only monitor the drive write-protect status every tick or so.
If it toggles, invalidate the cache.
I implemented it on a 8085 system, checking the write protect status
every 250 msec. The drive needn't be spinning. If you insert or remove
a 5.25" disk, the status will change.
We even went a bit further and checked to see if the disk contained any
files open for writing. If so, we sounded an alert and froze the system
until the subject disk was re-inserted. This reduced the number of
disk-swapping related errors essentially to zero.
And, it presumably worked just fine.
They tried TWO of their staff, and neither could change a disk in less
that 2 seconds.
COULD somebody go from write protected 8" or non-write-protected 5.25"
(did you sense door open?) to disk out of drive in under 250ms, and then
lso go from door open, no disk to non-write protected disk in drive (door
still open?) in under another 250ms? I can't. But if somebody
could, then you just need to change that user's settings to about 100ms.
4:34 p.m.
On 9/24/19 4:13 PM, Fred Cisin via cctalk wrote:
COULD somebody go from write protected 8" or
non-write-protected 5.25"
(did you sense door open?) to disk out of drive in under 250ms, and then
lso go from door open, no disk to non-write protected disk in drive
(door still open?) in under another 250ms??? I can't.? But if somebody
could, then you just need to change that user's settings to about 100ms.
It's impossible to flip open a drive and change floppies in less than a
couple of seconds. We tried. The drives were MIcropolis and MPI drives.
--Chuck
12:53 p.m.
On 9/24/2019 1:40 PM, Christian Liendo via cctalk wrote:
I meant to say Raymond Chen and it didn't notice
the spell check change
Sometimes modern technology sucks.
On Tue, Sep 24, 2019 at 3:18 PM Christian Liendo <cliendo at gmail.com> wrote:
Well I am
guessing, the time the flopppy motor was kept on after
a block read/write. Ben.
How did MS-DOS decide that two seconds was the amount of time to keep
the floppy disk cache valid?
https://devblogs.microsoft.com/oldnewthing/20190924-00/?p=102915
2 Oct
2 Oct
2:07 p.m.
New subject: Detecting inserted floppy drive WAS: Raymond Check discusses MS DOS Floppy Disk Cache
Just wondering if anyone (Chuck, Fred?) knows what is the exact technique referred to in
this other post by Raymond?
devblogs.microsoft.com/oldnewthing/20090402-00/?p=18643
-Ali
-----Original Message-----
From: cctalk [mailto:cctalk-bounces at classiccmp.org] On Behalf Of
Christian Liendo via cctalk
Sent: Tuesday, September 24, 2019 12:18 PM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Raymond Check discusses MS DOS Floppy Disk Cache
How did MS-DOS decide that two seconds was the amount of time to keep
the floppy disk cache valid?
https://devblogs.microsoft.com/oldnewthing/20190924-00/?p=102915
2:57 p.m.
New subject: Detecting inserted floppy drive WAS: Raymond Check discusses MS DOS Floppy Disk Cache
On Wed, 2 Oct 2019, Ali via cctalk wrote:
Just wondering if anyone (Chuck, Fred?) knows what is
the exact technique referred to in this other post by Raymond?
devblogs.microsoft.com/oldnewthing/20090402-00/?p=18643
As I understnd it, they just used a stopwatch and watched a couple of
their people change disks.
Since it was always 2 seconds or more, they used that as the number.
Chuck, on the other hand, polled the write protect switch every 1/4
second. If it changed state, he assumed disk change.
1/4 second assumes that nobody could take out an un-write protected 5.25
(or write protected 8") in less than 1/4 second, and then later put in a
different one in less than 1/4 second.
Unless you also have a way to monitor door opening (SOME drives), I have
an unnecessary worry that on an 8" write protected disk, you could change
disks. 'course, if the disk is write protected, then you don't need to
worry bout writing buffers to the wrong disk.
BTW, on "Computer Bowl" quiz show, NOBODY on Bill Gates' team could answer
the question of WHERE the write protect notch is on an 8" disk!
--
Grumpy Ol' Fred cisin at xenosoft.com
3:05 p.m.
New subject: Detecting inserted floppy drive WAS: Raymond Check discusses MS DOS Floppy Disk Cache
-----Original Message-----
From: cctalk [mailto:cctalk-bounces at classiccmp.org] On Behalf Of Fred
Cisin via cctalk
Sent: Wednesday, October 02, 2019 2:58 PM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Detecting inserted floppy drive WAS: Raymond Check
discusses MS DOS Floppy Disk Cache
On Wed, 2 Oct 2019, Ali via cctalk wrote:
Fred,
This is a different blog post. In this one he talks about how they found a
way to see if a disk was inserted in the floppy drive without spinning up
the motor....
Ali
Just wondering if anyone (Chuck, Fred?) knows
what is the exact
technique referred to in this other post by Raymond?
As I understnd it, they just used a stopwatch and watched a couple of
their people change disks.
Since it was always 2 seconds or more, they used that as the number.
3:15 p.m.
New subject: Detecting inserted floppy drive WAS: Raymond Check discusses MS DOS Floppy Disk Cache
Sorry, I was assuming the previous discussion!
This implies that some newer drives actually had a "disk present" signal,
or wouldn't let you close the door without a disk, but had a door signal.
Rather than an "unnecessary" training insertion, couldn't the testing of
drive type have simply waited until first actual use?
(a 5.25 would have shown write-protect while the disk was half in.)
Whether or not a drive had "DRIVE-READY" was not reliable enough, unless
you deprecated use of old drives, . . .
4:45 p.m.
New subject: Detecting inserted floppy drive WAS: Raymond Check discusses MS DOS Floppy Disk Cache
On 10/2/19 3:15 PM, Fred Cisin via cctalk wrote:
Sorry, I was assuming the previous discussion!
This implies that some newer drives actually had a "disk present"
signal, or wouldn't let you close the door without a disk, but had a
door signal.
Rather than an "unnecessary" training insertion, couldn't the testing of
drive type have simply waited until first actual use?
(a 5.25 would have shown write-protect while the disk was half in.)
Whether or not a drive had "DRIVE-READY" was not reliable enough, unless
you deprecated use of old drives, . . .
Regardless, the Win95 solution to identifying disk changes was to
attempt to rewrite the volume system ID and serial number in the boot
sector on every inserted disk that was write enabled. If that screwed
up the usability of the disk, well, tough luck.
They called it "volume tracking". You can identify such modified disks
as having the "xxxxxIHC" as the system ID, where "xxxxx" was some
random
value.
--Chuck
3:11 p.m.
New subject: Detecting inserted floppy drive WAS: Raymond Check discusses MS DOS Floppy Disk Cache
On 10/2/19 2:07 PM, Ali via cctalk wrote:
Just wondering if anyone (Chuck, Fred?) knows what is
the exact technique referred to in this other post by Raymond?
devblogs.microsoft.com/oldnewthing/20090402-00/?p=18643
Not enough information to know. It should be noted that one needn't spin
up drives to select them and poll them. I wonder if this is the
difference between "drive ready" or "disk changed" on pin 34 of the
interface.
-Chuck
1679
days inactive
1687
days old
12 comments
5 participants
participants (5)
-
bfranchuk＠jetnet.ab.ca -
cclist＠sydex.com -
cctalk＠ibm51xx.net -
cisin＠xenosoft.com -
cliendo＠gmail.com