:: Re: [DNG] FW: support for merged /u…
Page principale
Supprimer ce message
Répondre à ce message
Auteur: Rainer Weikusat
Date:  
À: dng
Sujet: Re: [DNG] FW: support for merged /usr in Debian
Didier Kryn <kryn@???> writes:

[...]

>>> I've read, from a guy who followed the story,that it was originally
>>> split because the first disk was too
>>> small.


[...]

>> There's a paper by Dennis Richtie, "The UNIX Time-Sharing System"


[...]

>>     The PDP-11 has a 1M byte fixed-head disk, used for file system
>>     storage and swapping, four moving-head disk drives which each
>>     provide 2.5M bytes on removable disk cartridges, and a single
>>     moving-head disk drive which uses removable 40M byte disk packs.

>>
>>          [...]

>>
>>     In our installation, for example, the root directory resides
>>     on the fixed-head disk, and the large disk drive, which con-
>>     tains user's files, is mounted by the system initialization
>>     program

>>


[...]

>> Which suggests functional reasons for the split: Keep the stuff
>> needed by everyone (and the swap space) on the fast disk and use the
>> slower one for 'individual users files'.
>
>     Good find Rainer. But I don't fully understand what you mean by:
>> Keep the stuff needed by everyone (and the swap space) on the fast disk
>> and use the slower one for 'individual users files'.

>
> Do you mean the applications in /usr/bin aren't used by everyone?
> Or they don't deserve to be launched quickly?


The main use of the original /usr was "store user home directories", ie,
files users were working with, including "less universally useful
programs".

> For me the only reason why they were not on the small fast disk is
> that this disk was full, ant it was full because it was small.
> Therefore, maybe Rob Landley is giving wrong details (or he refers to
> a later setup), but the reason he gives for the split makes full
> sense.


Looking somewhat further into available source, "The Evolution of the
Unix Time-sharing System" (Ritchie, 1979) says that

          During the last half of 1971, we supported three typists
    from the Patent department, who spent the day busily typing,
    editing, and formatting patent applications[*], and meanwhile
    tried to carry on our own work. Unix has a reputation for
    supplying interesting services on modest hard- ware, and this
    period may mark a high point in the benefit/equipment ratio; on
    a machine with no memory protection and a single .5 MB disk,
    every test of a new program required care and boldness, because
    it could easily crash the system, and every few hours' work by
    the typists meant pushing out more information onto DECtape,
    because of the very small disk.


        The experiment was trying but successful [...] we achieved
    sufficient credibility to convince our own management to acquire
    one of the first PDP 11/45 systems made. We have accumulated
    much hardware since then


There's also a "Notes for a UNIX talk ca 1972". Unfortunately, I don't
have a local copy of that, it's also not part of Landley's "sources" and
what used to be "Dennis Ritchie's homepage" is - at best -
intermittently accessible nowadays but I this 'UNIX(*) talk' refers to
the 11/45 installations and talks about "a fast, fixed-head system disk"
and a "large, slower, moving head disk [mounted at /usr]"

    where all the users' files are kept


Considering the expierence with using a shared, even smaller disk with
the first PDP-11 installation, it seems very probable that the 11/45 was
meant to be used with a "large" (and fairly cheap) "user disk" to begin
with, in order to get around the need to reload loads of stuff from tape
whenever a new user wanted to use the machine.

[*] 'Typing patent applications' using ed and nroff, doubtlessly happy
    with how user-friendly this system was when compared to a
    typewriter. Is is clerks or programmers who degenerated so much
    since then? :->>