Subject: Tops-10 filename/extension representation [was Re: Origin Of Filename Extension Dot Separator]

Newsgroups appropriately trimmed.

Johann 'Myrkraverk' Oskarsson <johann@myrkraverk.invalid> writes:

On 25/08/2026 12:20 PM, Mr. Chang Man-wai wrote:

On 8/25/2026 3:28 AM, Johann 'Myrkraverk' Oskarsson wrote:

This one is easy.ÿ You put the letter in an envelope, then put a name
one the envelope.ÿ Then you put the same name on a credit card sized
/index card/, and put the index card in the drawer that has the same
first letter as the name you put on the envelope in the first place.

That's just a library trick

But in the naming method,
which is the file name?
Which is the file name extension? :)

Yes, we have to wonder what's in a name. If I remember correctly, the
FAT [1] file system has separate entries for the /filename part/ and the /extension part/. And even if FAT doesn't have it, we can still imagine
a file system that does.

On disk, user file directories are made up of pairs of 36 bit words. The first word contains the SIXBIT (ASCII values reduced by 40(8)) name, up to 6 characters,
and the second word contains the SIXBIT extension in the left halfword and a pointer to the first block of the file in the right halfword. File attributes such as protection, write date, etc., are stored in that first block.

On DECtape, things are a little different.

1. The directory block is block 100(10) = 144(8) on the tape (out of a total of
578(10) blocks).
2. The first 83 words of the directory are broken into 5 bit bytes, with each
byte containing the file number (1 through 22(10)) to which that tape block
belongs.
3. The next 22 words contain the SIXBIT name of the file, up to six characters.
4. The next 22 words contain the SIXBIT extension, up to three characters, in the
left halfword, six 0 bits, and the file creation date (low order 12 bits). 5. The last word contains a tape label, the SIXBIT name of the tape.

The date format led to the DATE75 problem, because only 11 years (from the epoch
date of 1 January 1964) could be represented. A revised date model stored 3 high
order bits of date in the unused bits 35 of the first 83 words, interleaved by 22
(that is, file 1's high order bits were in words 0, 22, and 44, and so on).

The 83 words can represent more blocks than exist in 22 files. Blocks are reserved to a boot program, and get a "file number" of 36(8); blocks beyond 577 in
word 83 get a file number of 37(8).

--
Rich Alderson news@alderson.users.panix.com
Audendum est, et veritas investiganda; quam etiamsi non assequamur,
omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
--Galen

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Tops-10 filename/extension representation [was Re: Origin Of Filename Extension Dot Separator]

Rich Alderson <news@alderson.users.panix.com> writes:

Johann 'Myrkraverk' Oskarsson <johann@myrkraverk.invalid> writes:

Yes, we have to wonder what's in a name. If I remember correctly, the
FAT [1] file system has separate entries for the /filename part/ and the
/extension part/.

On disk, user file directories are made up of pairs of 36 bit words.
The first word contains the SIXBIT (ASCII values reduced by 40(8))
name, up to 6 characters, and the second word contains the SIXBIT
extension in the left halfword and a pointer to the first block of the
file in the right halfword.

I have this vague memory suggesting that RT-11 and RSX-11 suffered from
the same 6.3 FILNAM.EXT limitation, though using RADIX-50 instead of
SIXBIT. Was CP/M the system responsible for extending that to 8.3?

I don?t think anybody ever shipped a filesystem that tried to use
Baudot-Murray code for filenames...

Kragen

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Tops-10 filename/extension representation [was Re: Origin Of Filename Extension Dot Separator]

On Sat, 05 Sep 2026 10:14:01 -0300, Kragen Javier Sitaker wrote:

I have this vague memory suggesting that RT-11 and RSX-11 suffered from
the same 6.3 FILNAM.EXT limitation, though using RADIX-50 instead of
SIXBIT.

I think RSX-11 allowed 9.3 names.

I think the PDP-8 had an earlier operating system called OS/8 that
only allowed 2 characters for the extension.

Was CP/M the system responsible for extending that to 8.3?

Unix was already doing 14-character names before that.

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)