path: root/info/digests/nts-l/9402

========================================================================
Date:         Sat, 29 Jan 1994 02:12:06 -0500
Reply-To: NTS-L Distribution list <NTS-L@DHDURZ1.BITNET>
From: Laurent Siebenmann <laurent@MATH.TORONTO.EDU>
Subject:      extended features of Textures

extended features of Textures


 > Initex can load patterns (cumulative with TeX3)
 > only if
 >    1. no .fmt is loaded
 >    2. no paragraph has been typeset

Consulting five TeX implementations as to whether
restrictions 1. and 2. hold as stated, I find:

   Textures | OzTeX | Direct-TeX | mlTeX | C-TeX
1.   no       yes       yes         yes      yes
2.   no       no        no          no       no

I conclude that Bernd's correction to 2 is sustained
for standard TeX; and that my "correction" to 1.
is not --- as Peter Breitenlohner predicted.
I am very happy to be wrong here rather than in TUGboat.

   I was accurately describing *extended TeX* behavior
of Textures --- a subject certainly of interest here.

    I am surprised.  Hitherto I had not found any qualitative
deviation of Textures behavior from that of an enormous initex
with which one communicates mostly via Macintosh interface
tools.

    Has anyone else noticed further "extended" TeX features
of Textures? The obvious answer is that Textures has merged
initex into virtex.  But has it not just dropped
virtex?

    Extended TeXs aught to provide a complete catalog of their
deviations from the performance of standard TeX. (That
includes e-TeX.) Particularly so where ".tex" file
acceptabilityis involved.  Has Textures not done so?

    Laurent S

 *** "[Do not] put too much confidence in experimental results until
 *** they have been confirmed by theory."
 ***                      ---Arthur Stanley Eddington

========================================================================
Date:         Sat, 29 Jan 1994 02:15:15 -0500
Reply-To: NTS-L Distribution list <NTS-L@DHDURZ1.BITNET>
From: Laurent Siebenmann <laurent@MATH.TORONTO.EDU>
Subject:      e- for economy

e- for economy

 > IMHO if only a very, very small group of TeX users can not
 > switch to e-TeX because of very restricted hardware/software
 > limitations, we should tell them to get a better machine/a
 > better OS (and ignore them)...

If you want e-TeX users to be numerous, tell them that they
will be able to use as many modules of e-TeX as their
power and capacity and os permit.  And that, in TeX
compatibility mode, power and RAM requirements will be
comparable to that of TeX.

     Laurent S

========================================================================
Date:         Sat, 29 Jan 1994 02:14:51 -0500
Reply-To: NTS-L Distribution list <NTS-L@DHDURZ1.BITNET>
From: Laurent Siebenmann <laurent@MATH.TORONTO.EDU>
Subject:      Lessons from the past

Lessons from the past


1)  Textures has already erased the distinction between initex and
virtex. I hope e-TeX will do as much.

2)  Textures has already allowed pattern assimilation by
any format of TeX that does not contain a compacted hyphenation trie.
I hope e-TeX will do as much.  This allows one to quickly
(inter)nationalize precompiled formats.

3)  (ini)TeX 2 allowed one to erase the existing trie at any
time and input new patterns. I hope e-teX will do as much.
Even if it cannot be done on a language-by-language basis.
No new command is needed;  TeX could simply report that it
has been obliged to erase patterns.  Textures does not
currently have this feature.

    With (3), e-TeX could never be completely stopped
when a heavily multilingual job comes along --- say a
tourist brochure in seven languages.  At worst it would be
slowed.

    I see no reason why (1) and (2) would require one to

 > -- to remove the trie compaction in IniTeX,
 > -- change the code behind \pattern to build an already usable
trie

So this proposal not what Bernd is suggesting.  Are (1)(2)(3)
inadequate in practice?


         Laurent S

========================================================================
Date:         Mon, 31 Jan 1994 11:08:53 +0100
Reply-To: NTS-L Distribution list <NTS-L@DHDURZ1.BITNET>
From: Mariusz Olko <mariusz@THETA1.IFPAN.EDU.PL>
Subject:      Re: eTeX: Proposal for hyphenation patterns on demand
In-Reply-To:  <9401281933.AA03784@sigma.ifpan.edu.pl>

Tim Bradshaw writes:

 > More interestingly: just how big is the overhead of using initex rather
 > than virtex? How will it compare with the overhead from using, say
 > LaTeX3 rather than 2.09, which presumably people will be encouraged to
 > do.

For preprocessing patterns TeX needs a number of arrays size of which
can be astimated as
( 3*sizeof(int) + sizeof(char) +sizeof(boolean) ) * trie_size + 256*sizeof(int)

This gives about 120k for trie size of 30000.


Mariusz

========================================================================
Date:         Mon, 31 Jan 1994 11:51:18 +0100
Reply-To: NTS-L Distribution list <NTS-L@DHDURZ1.BITNET>
From: Mariusz Olko <mariusz@THETA1.IFPAN.EDU.PL>
Subject:      Re: eTeX: Proposal for hyphenation patterns on demand
In-Reply-To:  <9401282017.AA03813@sigma.ifpan.edu.pl>

Bernd Raichle writes:

 > When I wrote my reply to the proposal I thought about the possibility
 >   -- to remove the trie compaction in IniTeX,
 >   -- change the code behind \pattern to build an already usable trie and
 >   -- include the same code in VirTeX.
 >
 > This means that the trie data structures and routines have to be
 > changed and we can't compress the trie by identifying common subtries,
 > because a subtries will be propably changed...

We don't have to give up with compaction.  We can end up with partial
compaction ie.  when dumping format eTeX can compact whatever it has
ready and then each set can be compacted before use.  In current TeX a
flag is checked whether the table has already been compacted befor
typesetting is started.  It can be arranged that there is a separate
flag for each language.  It can be checked separately before
hyphenating word in current language and if this part of table is not
ready it can be compacted.

I have made test to see what is the overlap between the subtries of
Polish and English patterns.  Here are the results.
English patterns   6075 trie size as reported by iniTeX
Polish patterns    5505
      ------------------
       total      11580

When loading two sets together i get trie size 11344.  That means the
gain from compacting patterns together is 236 at the level of 2% !!!.
(Can anybody check this for other sets of paterns ?)

At the same time Ziri Zlatuska's experiments show that compacted
patterns are nearly 30% smaller than non compacted.


regards
                Mariusz

========================================================================
Date:         Tue, 1 Feb 1994 13:40:24 +0100
Reply-To: Chris Rowley <C.A.Rowley@open.ac.uk>
From: "Johannes L. Braams" <J.L.Braams@RESEARCH.PTT.NL>
Subject:      LaTeX2e conference announcement

               *******************************
               *   LaTeX2e: The Conference   *
               *******************************
                        ************
         Announcing the New Standard version of LaTeX
         ********************************************

You've read the hype, you've ordered the book---
                  --- haven't you? ---
          ---now here's the chance to discover the real thing!

For authors, publishers, support persons, ..., everyone---
         ---two days to discover what LaTeX2e can do for you

All the New Features of LaTeX2e---
            The New Standard Version of LaTeX
                          ---Incorporating SliTeX and AMSLaTeX
                        ************

  ***  how to install it

  ***  how to integrate your local styles

  ***  how to use scalable fonts

  ***  how to integrate graphics

  ***  how to process existing documents

  ***  how to use the new features

  ***  how to write extension packages

  Presentations by members of the LaTeX3 project team
  Featuring Frank Mittelbach, Johannes Braams, David Carlisle,
  Chris Rowley, Alan Jeffrey, Michel Goosens (at least)

***************************************************************
Booking Form

I wish to attend the UKTUG's LaTeX2e conference on March 21st--22nd:

Name:
Organisation:
Address:


Telephone:
Email:
Please specify any special dietary requirements:
Specify any changes to standard 1-night accommodation[2]:
Indicate whether you require a copy of `The \LaTeX\ Companion'[3]:

   Payment enclosed[1,4,9]
or purchase order number[5]
  Signed:
    Date:

Notes
[1] Payment for the conference, all meals (including a conference dinner),
     refreshments and accommodation for the night of Monday, March 21st,
     will be \pounds89 for members of  UKTUG (or other \TeX\ user groups)
     and \pounds120 for non-members.
[2] Additional nights' accommodation (bed and breakfast) can
     be included at a cost of \pounds20 per night.
     Indicate which additional nights you require
     (if any).  If you do not require accommodation for the night
     of Monday,  March 21st, deduct \pounds20 and note this.
[3] `The \LaTeX\ Companion', by  Goossens, Mittelbach  and Samarin,
     has recently been published by Addison Wesley. It is the main source of
     documentation about \LaTeX2e.  We hope to have this book
     available to delegates at a discount from its normal price of
     \pounds26.95. Please indicate whether you wish to take advantage of
     this offer, which is available only to those attending
     the conference.
[4] Sterling cheques should be made payable to `Warwick University'.
[5] A Purchase Order, with an order number, is an acceptable alternative
     method of payment, but will attract a \pounds15 surcharge on the
     amount due.
[6] If your cost is outside the range \pounds69--175, recalculate.
[7] Return the completed form to arrive by Tuesday March 1st, 1994.
     We cannot guarantee enrollment after that date. If you are unable to
     return the form by then, but are definitely coming, contact
     the local organiser by email, telephone, fax or mail to secure a place.
     For contact details, see below.
[8] If your requirements are not met by the options offered,
     contact the local organizer.
[9] Note that VAT is not chargeable.
[10] A proper LaTeX-able version of the booking form is available from
     the local organizer.

Meeting duration
Conference  registration  will commence at 1000 on  Monday 21st;
coffee  will be available from 1030; the sessions will start at 1115.
The sessions will end at 1700 on Tuesday 22nd.

Travel to Warwick University
There is an Inter-City rail service to Coventry, but from some
locations it may be easier to travel to Birmingham initially.

Journey                     depart  arrive
Euston to Coventry           0810    0920
                             0840    0950
                             0910    1025
Manchester to Coventry       0718    0929
                             0818    1027
Bristol (TM) to B'ham        0714    0854  Trains from Birmingham to Coventry
                             0830    0954  are frequent and fast (journey time
Newcastle to B'ham           0640    0951  of 20--30 minutes, departing
York to B'ham                0742    0951  every 10 or 15 minutes).

There will be a shuttle bus service running between 1000 and 1030 for delegates
arriving at Coventry Railway Station. The shuttle bus will also be available
on the 22nd to take delegates back to the station.

A taxi from the station to the University costs about \pounds4.
Alternatively, buses (notably the X10, X12, 12 and 12A) run from
just outside the station to the University, taking about 15 minutes.
It is a de-regulated service and as such impossible to guess the
fare from day to day. It is usually about 50p. The 12 and 12A services
are `correct fare only', but no-one (except the driver) knows what
the correct fare is.

The University is easily accessible by car, and
may be reached from the M1, M40, M42, M45, M6 and M69.
There is ample parking for private vehicles on campus.

Birmingham airport is about 12 minutes from Coventry by
rail and is connected by a frequent and fast service.

Payment Address
Please return the completed booking form, together
with your payment, to arrive before Tuesday March 1st, 1994, to:

Malcolm Clark (lxii)
Computing Services         Tel: 0203 523365
University of Warwick      Fax: 0203 523267
Coventry CV4 7AL         Email: m.clark@uk.ac.warwick (Janet)
UK                              m.clark@warwick.ac.uk (Internet)

                                                       Enrol by March 1st, 1994

========================================================================
Date:         Wed, 2 Feb 1994 13:46:51 +0100
Reply-To: NTS-L Distribution list <NTS-L@DHDURZ1.BITNET>
From: Peter Deussen <dem@CS.TU-BERLIN.DE>

unsubscibe

========================================================================
Date:         Fri, 4 Feb 1994 12:44:49 EST
Reply-To: NTS-L Distribution list <NTS-L%DHDURZ1.BITNET@vm.gmd.de>
From: "MARSOLAIS,SYLVAIN,MR" <BBFO@MUSICB.MCGILL.CA>

LIST GLOBAL

========================================================================
Date:         Thu, 17 Feb 1994 11:53:04 -0500
Reply-To: NTS-L Distribution list <NTS-L%DHDURZ1.BITNET@vm.gmd.de>
From: Michael Downes <MJD@MATH.AMS.ORG>
Subject:      Another debugging proposal

Proposed debugging extension:

---Improve the reporting of conditionals when \tracingmacros > 1 to make
it easier to locate the source of an error such as "File ended when \ifx
on line 1371 was incomplete". Since TeX already maintains the
information about the begin-conditional command and the line number, I
hope that it would not be too hard to write this into the log file at
the appropriate point, together with the current conditional nesting
level:

  {\ifx: 2 l.300}
  {false}
  {\fi: \ifx 2 l.300}

where "2" indicates the nesting level inside the conditional.

Motivation: There is currently no easy way to properly match up all the
\if's and \fi's in a tracing log, for reasons that are a little
complicated to explain. Anyone who has tried to debug this sort of error
in a complicated macro package will know what I'm talking about, I
think. The proposed extension would help in this way: If you want to
find the \ifx that matches a given \fi, you would merely have to search
backward for the pattern

  {\ifx: 2

The nesting level number guarantees that the first occurrence in the
reverse direction will be the proper match, provided that tracing was
turned on throughout the intervening material. The line number will help
track down a match when there is a large amount of intervening material
so that tracing through all of it would be inconvenient (for example, by
producing a log file that uses up all 7 megs of your remaining disk
space :-).

Michael Downes
mjd@math.ams.org