lpilindy@lindy.stanford.edu (Captain Beyond) (11/04/90)
In preparation to update Common TeX to version 3.0, i have made the following changes: 1. cleaned up strings, no string pool. i hope to extract all the strings soon and use a formatted print routine. but internally strings arent stored as counted vectors in a string pool. 2. no memory array, no "memorywords, halfwords, etc". C structures are used, with dynamic allocation. 3. restructured symbol table. no eqtb table, etc. symbols are still treated as "named registers", but the global registers dont share structure with the symbol table. 4. restructured font table. 5. simplified type system. 6. no virtex. these is only an initex, with undumped versions for tex, latex, etc. 7. all bug fixes to 2.95. actually beyond that, but i am waiting for 3.0 to distribute these. 8. ad nauseum.... there are more things on my list to do, but i think the current version is "stable". the changes are quite extensive, so i didnt bother keeping anyone up to date. hopefully this will no longer be the case. i dont think the changes to 3.0 will be as numerous as these. but i would like to get versions to people "interested" in steering Common TeX to 3.0. i am open to suggestions. i am also on the Internet, so i would value feedback from sites. portability needs to be addressed and i would rather do this with 2.95. obviously, using new C features will introduce shortcomings in my programming. please wait for version 3.0 for general inquiries for obtaining Common TeX. i am interested in reconnecting with TeX hackers for the moment. but i will make Common TeX 2.95 and Metafont 1.0 available for FTP access, if desired. please send messages to monardo@cshl.org. we dont yet have a news feed. but all mail will be answered. /pat
rusty@belch.Berkeley.EDU (rusty wright) (11/06/90)
Gack, you mean someone's still beating that dead horse? I mean, it's stiff; it's drawing flies! I can't believe it. Use web2c; it automatically generates the .c files from the web files so you get the newest versions automagically.
lpilindy@lindy.stanford.edu (Captain Beyond) (11/07/90)
people SHOULD use web2c if they just want to install it. Common TeX is my hobby and i hope to work on it for a long time. i dont enjoy working with translators so i have little choice. i am sure if i compared notes with people who worked with translators, i would find two things: 1. that they know more about the tranlator than TeX 2. that they didnt enjoy themselves. /pat
pcg@cs.aber.ac.uk (Piercarlo Grandi) (11/09/90)
On 5 Nov 90 19:47:04 GMT, rusty@belch.Berkeley.EDU (rusty wright) said: rusty> Gack, you mean someone's still beating that dead horse? I mean, it's rusty> stiff; it's drawing flies! I can't believe it. Use web2c; it rusty> automatically generates the .c files from the web files so you get the rusty> newest versions automagically. You must be joking of course. I'd rather have (and I am not alone in this) any time Monardo's Common TeX than Knuth's implementation of TeX (have you dared to read it?), to which your description above applies fairly well, especially if it is passed thru web2c (its embalmer). In order of increasing preference of TeX implementations *for Unix*: 1) The Knuth version passed thru web and compiled by a pascal compiler. 2) The Knuth version passed thru web2c and compiled by a C compiler. 3) The Knuth version passed thru web and p2c and compiled by a C compiler. Tolerable. Usually faster and easier to work with than either of the above two hacks. 4) Common TeX 2.93. Not bad, still twice as slow as troff. But smaller, simpler and more comprehensible and Unix portable than Knuth's version variously preprocessed. 5) Hopefully soon, Common TeX 2.95. Written in C using C as it should be, without fixed table sizes, memory arrays, etc... Should be as fast or faster than troff. Note that Common TeX stil has the defect that in order to pass the trip test is has to have the same grettable, verbose, ridiculous user interface as Knuth's TeX. As soon as I get Common TeX 2.95 I will put in a more Unixish user interface, conditionally compiled. -- Piercarlo Grandi | ARPA: pcg%uk.ac.aber.cs@nsfnet-relay.ac.uk Dept of CS, UCW Aberystwyth | UUCP: ...!mcsun!ukc!aber-cs!pcg Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@cs.aber.ac.uk
mcdonald@aries.scs.uiuc.edu (Doug McDonald) (11/11/90)
In article <PCG.90Nov9152114@odin.cs.aber.ac.uk> pcg@cs.aber.ac.uk (Piercarlo Grandi) writes: >On 5 Nov 90 19:47:04 GMT, rusty@belch.Berkeley.EDU (rusty wright) said: > >rusty> Gack, you mean someone's still beating that dead horse? I mean, it's >rusty> stiff; it's drawing flies! I can't believe it. Use web2c; it >rusty> automatically generates the .c files from the web files so you get the >rusty> newest versions automagically. > >You must be joking of course. I'd rather have So would I >(and I am not alone in >this) You got that right!!! >any time Monardo's Common TeX than Knuth's implementation of TeX >(have you dared to read it?), No!!! (PUKE RETCH!!) >to which your description above applies >fairly well, especially if it is passed thru web2c (its embalmer). > >In order of increasing preference of TeX implementations *for Unix*: > >1) The Knuth version passed thru web and compiled by a pascal compiler. I don't AHVE any Pascal compilers --- I don't need a lobotomy either. > >2) The Knuth version passed thru web2c and compiled by a C compiler. > I tried on two systems and never even got it close to starting to working. >3) The Knuth version passed thru web and p2c and compiled by a C >compiler. Tolerable. Usually faster and easier to work with than either >of the above two hacks. > >4) Common TeX 2.93. Not bad, still twice as slow as troff. But smaller, >simpler and more comprehensible and Unix portable than Knuth's version >variously preprocessed. I tried this on both Unix and <<MS-DOS>> and on Unix it compiled and ran just fine the very first time. On the PC I just had to fix the filename syntax, that is all. > >5) Hopefully soon, Common TeX 2.95. Written in C using C as it should >be, without fixed table sizes, memory arrays, etc... Should be as fast >or faster than troff. Anxiously waiting for this. It should totally obsolete the Knuth's version. > >Note that Common TeX stil has the defect that in order to pass the trip >test is has to have the same grettable, verbose, ridiculous user >interface as Knuth's TeX. As soon as I get Common TeX 2.95 I will put in >a more Unixish user interface, conditionally compiled. Please post when you're done - and please make sure it is not TOO Unix specific. Unix-ish in concept but not tied tightly to Unix itself would be perfect. >Piercarlo Grandi | ARPA: pcg%uk.ac.aber.cs@nsfnet-relay.ac.uk Thanks for the nice remarks. Doug McDonald
bjornl@sics.se (Bj|rn Lisper) (11/12/90)
In article <PCG.90Nov9152114@odin.cs.aber.ac.uk> pcg@cs.aber.ac.uk (Piercarlo Grandi) writes: >Note that Common TeX stil has the defect that in order to pass the trip >test is has to have the same grettable, verbose, ridiculous user >interface as Knuth's TeX. As soon as I get Common TeX 2.95 I will put in >a more Unixish user interface, conditionally compiled. This perhaps answers my inquiry here some time ago about the ability of TeX to pipe input and output in a unix environment. May I suggest that you make standard input for source and standard output for dvi default? (Note that this requires the error handling to be improved, so that TeX never can "get stuck" at a certain part of the text.) A simple mechanism to use run-time provided arguments from the command line would also be handy. And yes, it would also be nice to be able to kill a TeX process with ctrl-C. Just some suggestions. Bjorn Lisper
iwm@doc.ic.ac.uk (Ian Moor) (11/17/90)
In article <PCG.90Nov9152114@odin.cs.aber.ac.uk> pcg@cs.aber.ac.uk (Piercarlo Grandi) writes: > >Note that Common TeX stil has the defect that in order to pass the trip >test is has to have the same grettable, verbose, ridiculous user >interface as Knuth's TeX. As soon as I get Common TeX 2.95 I will put in >a more Unixish user interface, conditionally compiled. Don't make it conditionally compiled, make it a run time choice. I like explanatory error messages! You might consider ed's `?' as the ultimate error messsage, but I would like some more help. Being able to enter corrections if there is an error near the end of a long document is very convenient. By all means accept tex -r for rude (i.e. nothing but ?) and -t for terse messages. I don't want to criticize the author of Common TeX, but surely the same effort could have produced a good portable optimizing Pascal compiler for Unix, and thus all the tools written in Pascal, not just TeX. -- Ian W Moor ARPA: iwm@doc.ic.ac.uk JANET: iwm@uk.ac.ic.doc Department of Computing, Lecture: The transfer of information from Imperial College. the lecturer's notes to the students' 180 Queensgate without passing through the brains of London SW7 UK. either.