TUHS

tuhs@tuhs.org

12 participants
6467 discussions

by Dave Horsfall

Every so often I want to compare files on remote machines, but all I can do is to fetch them first (usually into /tmp); I'd like to do something like: rdiff host1:file1 host2:file2 Breathes there such a beast? I see that Penguin/OS has already taken "rdiff" which doesn't seem to do what I want. Think of it as an extension to the Unix philosophy of "Everything looks like a file"... -- Dave

9 months, 3 weeks

Re: [COFF] Re: Supervisor mode on ye olde PDP-11

by jnc＠mercury.lcs.mit.edu

> From: Warner Losh > 2.11BSD used a mode between kernel and user for the TCP stack to get > more effective address space... Is there a document for 2.11 which explains in detail why they did that? I suspect it's actually a little more complicated than just "more address space". The thing is that PDP-11 Unix had been using overlays in the kernel for quite a while to provide more address space. I forget where they first came in (I suspect there were a number of local hacks, before everyone started using the BSD approach), but by 2.9 BSD they were a standard part of the system. (See: https://minnie.tuhs.org/cgi-bin/utree.pl?file=2.9BSD/usr/src/sys/conf/Ovmak… for some clues about how this works. There is unfortunately no documentation that I know of which explains clearly how it works; if anyone knows of any, can you please let me know? Otherwise you'll have to read the sources.) I can think of two possible reasons they started using supervisor mode: i) There were a limited number of the 2.9-type overlays, and they were not large; trying to support all the networking code with the existing overlay system may have been too hard. ii) I think this one is unlikely, but I'll list it as a possibility. Switching overlays took a certain amount of overhead (since mapping registers had to be re-loaded); if all the networking code ran in supervisor mode, the supervisor mode mapping registers could be loaded with the right thing and just left. Noel

9 months, 3 weeks

Re: Version 256 of systemd boasts '42% less Unix philosophy' The Register

by jnc＠mercury.lcs.mit.edu

> From: Clem Cole > Frankly my major complaint with much of the modern world is that when we > ignore the past "There are two kinds of fools. One says, 'This is old, therefore it is good'; the other says, 'This is new, therefore it is better.'" -- Dean Inge, quoted by John Brunner in "The Shockwave Rider". Noel

9 months, 3 weeks

Diff: The seminal water we swim in

by Douglas McIlroy

> this month marks the fiftieth anniversary of the release of what > would become a seminal, and is arguably the single most important, > piece of social software ever created. I'm flattered, but must point out that diff was just one of a sequence of more capable and robust versions of proof(1), which Mike Lesk contributed to Unix v3. It, in turn, copied a program written by Steve Johnson before Unix and general consciousness of software tools. Credit must also go to several people who studied and created algorithms for the "longest common subsequence" problem: Harold Stone (who invented the diff algorithm at a blackboard during a one-day visit to Bell Labs), Dan Hirschberg, Tom Szymanksi, Al Aho, and Jeff Ullman. For a legal case in which I served as an expert witness, I found several examples of diff-type programs developed in the late 1960s specifically for preparing critical editions of ancient documents. However, Steve Johnson's unpublished program from the same era appears to be the first that was inspired as a general tool, and thus as "social software". Doug

9 months, 3 weeks

Diff: The seminal water we swim in

by Rich Salz

From Mastodon, MHoye posting at https://mastodon.social/@mhoye/112615649865163136: I haven't seen anybody mentioning it or even noticing it, like it's just the water we swim in now, but this month marks the fiftieth anniversary of the release of what would become a seminal, and is arguably the single most important, piece of social software ever created. Written by Douglas McIlroy and James Hunt and released with the 5th Edition of Unix this month in 1974: diff. https://minnie.tuhs.org/cgi-bin/utree. <https://minnie.tuhs.org/cgi-bin/utree.pl?file=V5/usr/source/s1/diff1.c>..

9 months, 3 weeks

5ESS UNIX RTR Reference Manual - Issue 10 (2001)

by segaloco

Good evening, while I'm still waiting on the full uploads to progress (it's like there's a rule any >100MB upload to archive.org for me has to fail like 5 times before it'll finally go...) I decided to scrape out the UNIX RTR manual from a recent trove of 5ESS materials I received and tossed it up in a separate upload: https://archive.org/details/5ess-switch-unix-rtr-operating-system-reference… This time around I've got Issue 10 from December 2001. The last issue of this particular manual I found on another 5ESS disc is Issue 7 from 1998 which I shared previously (https://ia601200.us.archive.org/view_archive.php?archive=%2F12%2Fitems%2F5e…) The manual is in "DynaText" format on the CD in question, unlike Issue 7 which was already a PDF on its respective CD. I used print-to-PDF to generate the above linked copy. Given that the CD itself is from 2007, this may point to UNIX RTR having no significant user-visible changes from 2001 to 2007 that would've necessitated manual revisions. In any case, I intend to upload bin/cue images of all 7 of the CDs I've received which span from 1999 to 2007, and mostly concern the 5ESS-2000 switch from the administrative and maintenance points of view. Once I get archive.org to choke these files down I also intend to go back around to the discs I've already archived and reupload them as proper bin/cue rips. I was in a hurry the last time around and simply zipped the contents from the discs, but aside from just being good archive practice, I think bin/cue is necessary for the other discs as they seem to have control information in the disc header that is required by the interactive documentation viewers therein. All that to say, the first pass will result in bin/cues which aren't easily readable through archive.org's interface, but I intend to also swing back around on these new discs and provide zips of the contents as well to ensure the archives are both correct (bin/cue) and easily navigable (zip). As always, if you have any such documentation or leads on where any may be awaiting archival, I'm happy to take on the work! - Matt G.

10 months

Re: [tuhs] Early statistical software in Unix

by Nelson H. F. Beebe

Doug McIlroy kindly sent me contact information for John Chambers, co-author of the cited book about the S system. I have just heard back from John, who offered a link to his summary paper from the 2020 HOPL conference proceedings S, R, and data science https://doi.org/10.1145/3386334 and reported that S was licensed to AT&T Unix customers only in binary form, and that the original source code may no longer exist. That is a definitive answer, even if not the one that I was hoping to find. ------------------------------------------------------------------------------- - Nelson H. F. Beebe Tel: +1 801 581 5254 - - University of Utah - - Department of Mathematics, 110 LCB Internet e-mail: beebe(a)math.utah.edu - - 155 S 1400 E RM 233 beebe(a)acm.org beebe(a)computer.org - - Salt Lake City, UT 84112-0090, USA URL: http://www.math.utah.edu/~beebe/ - -------------------------------------------------------------------------------

10 months

Re: most direct Unix descendant

by Douglas McIlroy

Eloquently put. Amen! doug > Unix brought automation to the forefront of possibilities. Using Unix, > anyone could do it - even that kid in Jurassic Park. Today, everything > is GUI and nothing can be automated easily or, most of the time, > not at all. > Unix is an ever shrinking oasis in a desert of non-automation and complexity. > It is the loss of automation possibilities that frustrates me the most

10 months

Draft: London and Reiser's UNIX/32V paper, reconstructed

by G. Branden Robinson

Hi folks, Reiser and London's paper documenting their preparation of UNIX/32V, a port of Seventh Edition Unix to the VAX-11/780, is an important milestone in Unix development--as much I think for its frank critique of C as "portable assembly" as for the status of the system documented: the last common ancestor of the BSD and System V branches of development. Because the only version I've ever seen of this paper is a scan of, possibly, a photocopy several generations removed from the original, I thought I'd throw an OCR tool at it and see about reconstructing it, not just for posterity but to put the groff implementation of mm to the test. So even if someone has a beautiful scan of this document elsewhere, this exercise remains worthwhile to me for what it has shown me about Documenter's Workbench mm and groff's mostly compatible reimplementation thereof. Please find attached my first draft of the reconstruction as an mm document as well as a PDF rendered by bleeding edge groff. I did not attempt to fix _any_ typos, solecisms, or non-idiomatic *roff usage (like the employment of hyphens as arithmetic signs or operators) in this document. I may have introduced errors, however, due to human fallibility or incorrect inferences about what lay beneath scanning artifacts. Hence its draft status. I welcome review. Assuming this reconstruction survives peer scrutiny, I aim to put it up on GitHub as I did Kernighan & Cherry's "Typesetting Mathematics" paper.[1] For the casual reader, I extract my documentary annotations below. For groff list subscribers, I will add, because people are accustomed to me venturing radical suggestions for reforms of macro packages, I suggest that we can get rid of groff mm's "MOVE" and "PGFORM" extensions. They're buggy (as the man page has long conceded), and I don't think anyone ever mastered them, not even their author. I rewrote "0.MT", essential to rendering of this document, without requiring them at all. I _tried_ to use them, but "MOVE" in particular introduced baffling errors in vertical spacing. When I threw it aside to attack head-on the layout problems facing me, things got easier. Further, simple caching and restoration of `.i` and `.l` register values (when multiple changes were being made to them within a macro) obviated `PGFORM`. I'm not sure that it is tractable to idiot-proof manipulations of basic layout parameters like these, as these macros seem to have tried to do. If a document author wants to seize control of page layout from a full-service macro package and reach deep into the guts of the formatter, they should glove up and put things back where they found them. My opinion. .\" London & Reiser's UNIX/32V porting paper .\" .\" Reconstruction in groff mm (but DWB 3.3 mm compatible) .\" from scanned/OCRed text by G. Branden Robinson, June 2024 .\" .\" The original scan shows no evidence of superscript usage, except on .\" the cover sheet where "TM" superscripts "UNIX". .\" .\" Some differences may arise due to changes in the mm macro package .\" itself from its PWB incarnation (ca. 1978) and DWB 3.3 (July 1992). .\" Thanks to Dan Plassche for the history. .\" https://www.tuhs.org/pipermail/tuhs/2022-March/025545.html .\" .\" The groff reimplementation of mm was undertaken mostly from .\" 1991-1999 (by Juergen Haegg), based on the DWB documentation. It .\" added features but also parameterized many aspects of package .\" behavior, for example to facilitate easy localization. Later, .\" Werner Lemberg and G. Branden Robinson contributed enhancements, bug .\" fixes, and improvements to the groff_mm(7) man page. .\" .\" I anticipate adding further parameters to groff mm to better .\" emulate the old version of mm used by this paper. (For example, the .\" format of the caption applied to the reference page differs between .\" PWB mm and DWB 3.3.) Where this document exercises such extensions, .\" they should be prefixed with a `do` request so that AT&T troff will .\" ignore them. .\" Override: "By default, ... bold stand-alone headings are printed .\" in a size one point smaller than the body." .\" XXX: The cover "page" (more like a header block) is a mess when .\" typeset with groff mm, and outright horrific in nroff mode. GBR has .\" fixes for these pending for push to GNU Savannah's Git repository. .\" .\" XXX: Original scan capitalizes "Subject:"; DWB 3.3 renders it in .\" full lowercase. .\" .\" XXX: Original scan bears a "TM:" heading for the technical .\" memorandum number(s). DWB 3.3 lacks this. .\" .\" Memorandum captions may have changed from PWB to DWB 3.3 mm. groff .\" mm has changed in Git (June 2024) to use the captions documented in .\" the DWB 3.3 manual. We override the default for authenticity. .\" XXX: Original scan sets reference marks as a typewriter might, at .\" normal size on the baseline between square brackets. DWB 3.3 .\" converts them to superscripts but keeps the brackets(!). groff mm .\" should add a "Rfstyle" register to control this. .\" 0 = auto (nroff/troff); 1 = bracket; 2 = superscript; 3 = both. (?) \" straight quotes in original .ns \" XXX: Hack to keep `VL` from adding pre-list vertical space. \" recte: *(\-\-p+i) \" bad ellipsis spacing in original \" - missing; error in text or scanner fubar? \" recte: \-1 \" sic .\" Either `AL` worked differently in 1978 mm, or didn't exist, or .\" somebody wanted this list _just so_. .\"AL "" 5 .\" XXX: Scan has signatures set farther to the right, not centered as .\" DWB 3.3 mm sets them. groff mm follows DWB here. .\" .\" XXX: PWB and DWB 3.3 put the signature names in bold; groff mm sets .\" them at normal weight. Bug. .\" .\" XXX: Scan has a couple of vees between the signature line and the .\" flush left secretarial annotation. groff mm sets the annotation on .\" the same line as the last author but also puts its information in .\" the cover page header as DWB 3.3 does, described next. DWB 3.3: (1) .\" omits the secretarial annotation altogether, putting it up in the .\" cover page header under the authors' names; (2) does not use author .\" initials (in the cover header) for this memorandum type; (3) puts .\" the department number after "Org." on the line under the author .\" name; (4) puts the abbreviated AT&T site name below that. Should we .\" consider a `Sgstyle` register for groff mm? .\" .\" XXX: groff mm organizes the department and site name differently .\" from DWB 3.3 in the cover head, and I don't see any reason for it .\" to. Fix this. .\" XXX: Scan only breaks between notations; DWB 3.3 and groff put 1v .\" between them. Should we consider an `Nss` register for groff mm? .\" XXX: Scan has references caption set flush left, in mixed case and .\" bold (just like `HU`). DWB 3.3 and groff center it and set it in .\" full caps in italics (at normal weight). If there were a way to .\" dump the accumulated reference list independently of rendering the .\" caption, that would give the author much more flexibility. .\" .\" XXX: The numbered reference list does not look like one produced .\" with `RL` nor with `AL`. The numeric tag is left-aligned within the .\" paragraph indentation. groff mm aligns it to the right. .\" .\" DWB 3.3 and Heirloom mm don't seem to honor `.RP 2` as the DWB .\" manual documents. They start the table immediately after the .\" reference list and go haywire boxing the table. Bug. Regards, Branden [1] https://github.com/g-branden-robinson/retypesetting-mathematics

10 months

PID 0 and the scheduler

by Will Senn

All, There's an interesting dive into PID 0 linked to from osnews: https://blog.dave.tf/post/linux-pid0/ In the article, the author delves into the history of the scheduler a bit - going back to Unix v4 (his assembly language skills don't go to PDP variants). I like the article for two reasons - 1) it's clarity 2) it points out the self-reinforcing nature of our search ecosystem. I'm left with the question - how did scheduling work in v0-v4? and the observation that search really sucks these days. Later, Will

10 months

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

TUHS