All the way up to 2004 or 2005, my home router was an old 486dx box running FREESCO (https://freesco.info/) - and it was, indeed, booting from a single floppy. Linux 2.0.something.
To my surprise I discovered today that FREESCO was still releasing updates all the way until 2014.
As a purveyor of floppy disk technology including MFM copiers, SCSI FDs, PS/2 FDs, and NOS ED disks and FDs, I can say that floppies suck and should be assumed to be bad and that any success of reading/writing/formatting is indeed a miracle. Any reading or writing of them will surely lead to their failure soon. And, they will fail just sitting there too.
Back in the day, I did try inducing errors into 3.5" HD floppies by waving a cheap, small voice-coil speaker magnet over a floppy and comparing its contents sector-by-sector to a disk image stored previously. I was only able to do this after opening the shutter and touching the magnet directly to the diskette surface.
PSA: Never write to 5.25" 360K floppies using 1.2M drives because the created track width is too small to be picked up by 360K drives.
Two decades ago, my former manager saved the day after a university hardware firewall went down.
Over the phone, he instructed the guy to download this floppy disk firewall, specify the ports, traffic direction, basic stuff.
Boot the computer and voila, the university network was up again giving the teams hours to get things sorted out.
I forgot its name now but many business would run this floppy disk firewall coz it just worked, once booted it would run "forever".
Interesting history, which does seem to be true: In the Beginning, the reliability of floppies was really quite good. Over time, as the mass market developed, however, quality went down as people shaved pennies and fractions of pennies. Storied brands such as Sony ended up being bad sector magnets (well, you know what I mean...).
My big secret for those of us still using floppies (USB floppies are a steal!) --- get Maxell NOS; they're about $3 per diskette now, but I have beat them hard and they keep going. Of course, I also had a batch of poorly-stored floppies that would not even turn (!) from one vendor (cheap, tho...) and they almost ruined one drive.
My rule now is: once a single sector goes kaput, make a new working copy from your backup (you HAVE a backup, right???), mark the bad sector of the failed disk, and put it in the 'BAD' pile. I have not yet had a sector on a Maxell 1.44 diskette go bad. Several SONY diskettes however are in that pile.
Note that the head drags on the surface of a floppy, so you're always losing a little; what that closeness does, tho, is allow phenomenal bits-per-square-inch. The coatings are very good, binders are good, magnetic properties are excellent. And the head widths (both sides, after all) are small. So you get formatted 80 cylinders of 18,432 bytes per cylinder MSDOS-compatible capacity, at that 5 revolutions per second ( 300 RPM, but gosh, RPS is a better unit here...) speed.
The floppy drive is an incredible chunk of engineering. It's unfortunate that Zip 100 and Zip 250 arrived so relatively late in the game.
I was partial to Verbatim because those seemed to be pretty consistently good from the early 80's through the mid 90's.
Zip drives sold like absolute mad mid/late 1995.*
* In 95, I placed around the top for the high school virtual portfolio investing contest ROI with 90% in IOMG +150% (2.5x) in 2 weeks. I also happened to be selling Zip drives at my night/weekend job as fast as they came in. My high school econ teacher mentioned his investing club invested and sold it before sales took off stonks-style and so they lost money on it. These floptical/-alikes went poof as soon as CD+/-R(W) drives and discs came down in price.
In my teens I ran a combination ipmasq(NAT, this was back when we called it ip masquerading) firewall and dial-up POP for my girlfriend at the time off a scrap 386 motherboard some ISA NICs and a 3.5" 1.44MB floppy drive. It was packed full of SIMMs, I don't remember how much probably 8MB RAM.
The userspace was all in the initramfs, linux booted directly without any LILO or GRUB (this was back in the days the kernel included its own boot loader), and the floppy drive was totally out of the picture once the system was up and running from RAM.
Prior to adding the dial-up aspect for my gf to share my internet from her home, the init was deliberately exited which technically panicked the kernel. Basically it was a /linuxrc shell script setting up the networking then deliberately breaking userspace - not even PID1 existed while it was just my firewall. The kernel keeps doing networking stuff even if panicked.
Iomega's awesome Zip drive disk (100MB, 250MB, 750MB capacities) , I think I still have a 250MB zip drive somewhere in my home attic.
They required a dedicated zip drive (took up same sized slot/bay as a floppy disk drive), but (if I recall right) that drive was backward compatible standard 3&1⁄2-inch 1.44MB floppy disks.
Interestingly, these drive also came in variants to work with different types of interfaces: IDE, ATAPI, USB, SCSI, FireWire.
Zip drives filled the portable storage niche, until CDs and DVDs replaced their need.
I found it cool that floppies and superfloppies had label stickers on which we can write (with a sketch pen) to remind the user of what content the disk is intended for.
There were some nice cameras that used Zip disks for storage! Very convenient for photographers working on multiple projects or sessions.
> They required a dedicated zip drive (took up same sized slot/bay as a floppy disk drive), but (if I recall right) that drive was backward compatible standard 3&1⁄2-inch 1.44MB floppy disks.
Zip was a completely unique physical format, and had no backwards compatibility with standard 3½" disks.
SuperDisk, on the other hand (in both the LS-120 and LS-240 variants) was backwards compatible with standard floppy disks in the same drive.
RX-01 DEC / IBM 3740 compatible was 77 tracks, single-sided, 128 bytes per sector and 26 sectors per track. Total 256,256 bytes. FM Modulation. 360 RPM. Disk to drive buffer: 4 µsec per data bit. Track-to-Track Seek: 6 ms. Head Settle Time: 25 ms. Average Access: Approximately 262 ms. 8" diameter diskette
One of these was used to load the microcode into the VAX-11/780 upon boot.
My old PDP11/73 (now in a museum) had two RX02, never had an RX01. Surprisingly fast! It also had two RL02s and a couple of RD54s in.
Building an RT11 system disk onto an RL02 off another RL02 made the downstairs neighbours complain quite a lot, even though the floor slab in my flat was about 40cm thick concrete. They didn't muck about with these 1960s tower blocks but it was no match for a pair of pint glass sized head actuators and a pair of washing machine motors.
In case the single floppy disk running everything fails: https://web.archive.org/web/20260129015513/http://floppy.ddn...
Watching the video now! Your stuff all rules btw
All the way up to 2004 or 2005, my home router was an old 486dx box running FREESCO (https://freesco.info/) - and it was, indeed, booting from a single floppy. Linux 2.0.something.
To my surprise I discovered today that FREESCO was still releasing updates all the way until 2014.
As a purveyor of floppy disk technology including MFM copiers, SCSI FDs, PS/2 FDs, and NOS ED disks and FDs, I can say that floppies suck and should be assumed to be bad and that any success of reading/writing/formatting is indeed a miracle. Any reading or writing of them will surely lead to their failure soon. And, they will fail just sitting there too.
Back in the day, I did try inducing errors into 3.5" HD floppies by waving a cheap, small voice-coil speaker magnet over a floppy and comparing its contents sector-by-sector to a disk image stored previously. I was only able to do this after opening the shutter and touching the magnet directly to the diskette surface.
PSA: Never write to 5.25" 360K floppies using 1.2M drives because the created track width is too small to be picked up by 360K drives.
Two decades ago, my former manager saved the day after a university hardware firewall went down.
Over the phone, he instructed the guy to download this floppy disk firewall, specify the ports, traffic direction, basic stuff. Boot the computer and voila, the university network was up again giving the teams hours to get things sorted out.
I forgot its name now but many business would run this floppy disk firewall coz it just worked, once booted it would run "forever".
Could QNX do this?
Seems to be timing out for me (not necessarily surprising, given that it is a floppy disk and HN has a history of hugging sites to death)
Archive link incase people want to see the page but can't load it: https://web.archive.org/web/20260129015513/http://floppy.ddn...
Perhaps no floppy drive has ever had so many patiently waiting for it to complete its work.
If you like this you'll love the web server on tape. Navigation forward works great but if you hit the back button it takes a while to rewind.
It would be even longer of a wait if it stopped to play Floppotron renditions of music for web users as analog audio streams.
Around year 2000 I had a camera that stored its images on a floppy disk.
Funny enough, that's the camera OP used to take the pictures in the floppy-disk hosted blog!
I'm a big fan OP, cool project!
an Apple camera perhaps? I loved that camera
Or a Sony Mavica, I love those
That is actually the one used to take picture of said computer and floppy drive on the website.
Anyone other than me remember the Cisco Micro Web Server?[1].
[1] https://www.employees.org/univercd/Feb-1998/cc/td/doc/prod_c...
Hope the hn post doesn't result in a bad sector on this floppy.
Interesting history, which does seem to be true: In the Beginning, the reliability of floppies was really quite good. Over time, as the mass market developed, however, quality went down as people shaved pennies and fractions of pennies. Storied brands such as Sony ended up being bad sector magnets (well, you know what I mean...).
My big secret for those of us still using floppies (USB floppies are a steal!) --- get Maxell NOS; they're about $3 per diskette now, but I have beat them hard and they keep going. Of course, I also had a batch of poorly-stored floppies that would not even turn (!) from one vendor (cheap, tho...) and they almost ruined one drive.
My rule now is: once a single sector goes kaput, make a new working copy from your backup (you HAVE a backup, right???), mark the bad sector of the failed disk, and put it in the 'BAD' pile. I have not yet had a sector on a Maxell 1.44 diskette go bad. Several SONY diskettes however are in that pile.
Note that the head drags on the surface of a floppy, so you're always losing a little; what that closeness does, tho, is allow phenomenal bits-per-square-inch. The coatings are very good, binders are good, magnetic properties are excellent. And the head widths (both sides, after all) are small. So you get formatted 80 cylinders of 18,432 bytes per cylinder MSDOS-compatible capacity, at that 5 revolutions per second ( 300 RPM, but gosh, RPS is a better unit here...) speed.
The floppy drive is an incredible chunk of engineering. It's unfortunate that Zip 100 and Zip 250 arrived so relatively late in the game.
RAID-sneaker with human-attended repair ;)
I was partial to Verbatim because those seemed to be pretty consistently good from the early 80's through the mid 90's.
Zip drives sold like absolute mad mid/late 1995.*
* In 95, I placed around the top for the high school virtual portfolio investing contest ROI with 90% in IOMG +150% (2.5x) in 2 weeks. I also happened to be selling Zip drives at my night/weekend job as fast as they came in. My high school econ teacher mentioned his investing club invested and sold it before sales took off stonks-style and so they lost money on it. These floptical/-alikes went poof as soon as CD+/-R(W) drives and discs came down in price.
Hopefully the floppy is read only at boot time and we hit the copy on ram.
In my teens I ran a combination ipmasq(NAT, this was back when we called it ip masquerading) firewall and dial-up POP for my girlfriend at the time off a scrap 386 motherboard some ISA NICs and a 3.5" 1.44MB floppy drive. It was packed full of SIMMs, I don't remember how much probably 8MB RAM.
The userspace was all in the initramfs, linux booted directly without any LILO or GRUB (this was back in the days the kernel included its own boot loader), and the floppy drive was totally out of the picture once the system was up and running from RAM.
Prior to adding the dial-up aspect for my gf to share my internet from her home, the init was deliberately exited which technically panicked the kernel. Basically it was a /linuxrc shell script setting up the networking then deliberately breaking userspace - not even PID1 existed while it was just my firewall. The kernel keeps doing networking stuff even if panicked.
Fun times.
> linux booted directly without any LILO or GRUB (this was back in the days the kernel included its own boot loader)
You can still do that if your machine supports coreboot.
> this was back in the days the kernel included its own boot loader
I mean hey these days you dont necessarily need some external bootloader either, on uefi systems the kernel can boot itself thanks to the efi stub.
I also remember this amazing tiny thing, tomsrtbt. It was packed full of tools, and also had an http server in it.
[dead]
That was some fast video editing!
720 kb SD or 1.44 MB DD ?
:-D
Sorry, nitpick: 720kB DD 1.44MB HD
It's funny that we've always called them "1.44 MB" disks when they actually hold 1440 KiB, which is 1.41 MiB or 1.47 MB.
"1.44" is a horrible mix of binary kilo and decimal mega which makes no sense.
With the GNU Units program, I have this defined in my ~/.units: "floppyMB 1000 KiB"
Is it useful? Perhaps not, but you can use it to translate "1.2 floppyMB", "1.44 floppyMB" into other units.
Probably because they have 2880 sectors.
++1
:))
One could cheat and run a SuperDisk drive, with 120 MB! Though that's not in the spirit of the game.
but do you remember ZIP-drives? :-) (was that the name?)
Yes! Superfloppy!
Iomega's awesome Zip drive disk (100MB, 250MB, 750MB capacities) , I think I still have a 250MB zip drive somewhere in my home attic.
They required a dedicated zip drive (took up same sized slot/bay as a floppy disk drive), but (if I recall right) that drive was backward compatible standard 3&1⁄2-inch 1.44MB floppy disks.
Interestingly, these drive also came in variants to work with different types of interfaces: IDE, ATAPI, USB, SCSI, FireWire.
Zip drives filled the portable storage niche, until CDs and DVDs replaced their need.
https://en.wikipedia.org/wiki/Zip_drive
I found it cool that floppies and superfloppies had label stickers on which we can write (with a sketch pen) to remind the user of what content the disk is intended for.
There were some nice cameras that used Zip disks for storage! Very convenient for photographers working on multiple projects or sessions.
https://www.digitalkameramuseum.de/en/prototypes-rarities/it...
> They required a dedicated zip drive (took up same sized slot/bay as a floppy disk drive), but (if I recall right) that drive was backward compatible standard 3&1⁄2-inch 1.44MB floppy disks.
Zip was a completely unique physical format, and had no backwards compatibility with standard 3½" disks.
SuperDisk, on the other hand (in both the LS-120 and LS-240 variants) was backwards compatible with standard floppy disks in the same drive.
ZIP Drives died as Minidisc died: MD was a very proper medium, but the inventors made some wrong decissions
I have a bunch of them, some SCSI ones in old samplers.
Yes. Along with the feared click of death.
Hey, my oscilloscope has one of those!
720k? In my day floppy disks had 96K and we liked it!
I liked when floppy disks were actually floppy.
3.5" floppies are still floppy. The case may be hard, but the floppy flops.
this was 8.25" back then?
RX-01 DEC / IBM 3740 compatible was 77 tracks, single-sided, 128 bytes per sector and 26 sectors per track. Total 256,256 bytes. FM Modulation. 360 RPM. Disk to drive buffer: 4 µsec per data bit. Track-to-Track Seek: 6 ms. Head Settle Time: 25 ms. Average Access: Approximately 262 ms. 8" diameter diskette
One of these was used to load the microcode into the VAX-11/780 upon boot.
My old PDP11/73 (now in a museum) had two RX02, never had an RX01. Surprisingly fast! It also had two RL02s and a couple of RD54s in.
Building an RT11 system disk onto an RL02 off another RL02 made the downstairs neighbours complain quite a lot, even though the floor slab in my flat was about 40cm thick concrete. They didn't muck about with these 1960s tower blocks but it was no match for a pair of pint glass sized head actuators and a pair of washing machine motors.