LATEX-L Archives

Mailing list for the LaTeX3 project

LATEX-L@LISTSERV.UNI-HEIDELBERG.DE

Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Condense Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Sender:
Mailing list for the LaTeX3 project <[log in to unmask]>
Subject:
From:
Frank Mittelbach <[log in to unmask]>
Date:
Mon, 23 Jun 1997 21:25:18 +0200
Reply-To:
Mailing list for the LaTeX3 project <[log in to unmask]>
Parts/Attachments:
text/plain (254 lines)
   [.... continued from previous mail ]

 > [from SPQR]
 >
 > >> i am sure that Phil, for instance,
 > >> would be the first to agree that what e-TeX needs very much is use and
 > >> criticism of their work.
 >
 > Agreed.  Of course, along with the criticism, a little praise now and then
 > wouldn't go amiss :-)

i guess you are aware that i wouldn't spend my time sitting sunday
evening in front of my computer with a bandaged hand and writing mails
like this if i wouldn't think it worth while --- see it as praise for
the work your team has done.

 > [from Chris Rowley]
 >
 > >> > switching eTeX to Unicode would be nice...
 > >>
 > >> If meaningful:-)!
 >
 > Quite!

red herring. unicode being short for a large enough unique internal
coding and you know it. please don't use Chris' comment to make it appear as
there is no point in that approach.

(see proceedings of the 10th Unicode conference with a paper by Chris
and me where we question certain aspects of unicode if you want to
interpret that statement :-)


 > >> I recall that the last discussion I had (with Phil) on this point
 > >> ended with us agreeing that it was probably not necessary to add
 > >> anything to the LaTeX kernel to support its use with e-TeX.
 >
 > >> Thus there appears to be no need to "support the new primitives in LaTeX".
 >
 > I agree completely with Chris that there is "no need"; but it seems to me
 > that the question is not one of need but of philosophy.  The philosophy
 > of LaTeX is to conceal the TeX typesetting system from the user, making
 > it hard (indeed, sometimes impossible) to achieve things in the "TeX"
 > manner, and forcing the user to achieve the same thing in "the LaTeX manner".

ouch. on the whole i strongly disagree with a statement like this but
this is no here and there

 > It is therefore surely not sensible to make an exception to this, and
 > effectively to say "we have hidden as much of TeX from you, the user, as
 > we possibly could; however, please feel free to use any and all of the new
 > e-TeX primitives", as if they were in some way different.  The new primitives
 > are there primarily to help people such as the LaTeX team: \protected,
 > \unexpanded and their ilk are not for Joe User, they are for format writers.

right and that is why i do welcome them so much in theory. and i agree it is not
for Joe User (or was it Ben User?) to use them directly but i think i've said often
enough why right now we think we can't integrate them (unfortunately) into the
kernel directly.

 > So although I agree with Chris that there is _no need_ for LaTeX to take
 > advantage of the e-TeX primitives, and that Joe User can use any and all
 > that he/she chooses, it doesn't somehow square with the LaTeX philosophy.
 > Chris?

true again but Joe package writer could and that would be helpful in
the current situation unless it resolves into what i consider the
nightmare of the "best typeset with ..." competitions.


 > [from Frank Mittelbach]
 >
 > >> it will but we are still talking about a small number and too small
 > >> without any additional boost that also brings us into all the non-unix
 > >> worlds
 >
 > There's only one platform for which e-TeX is still unavailable, of the
 > major platforms in use today: the Apple Macintosh.  Versions exist for
 > all other major platforms, so there's no real obstacle to its uptake
 > other than inertia...

there is a big difference between theoretical and practical
availability. it has to be in *all* or at least most of the major
public distributions and also important it has to be in the commercial
ones. since one end of the TeX community (the publishers and other
companies) tend to use the commercial distributions this is also an
important aspect.

 > >> this is why for formats like lollipop nobody ever wrote additional
 > >> modules, not because it is a bad product, far from it, but because it
 > >> didn't have the backing and that killed it, or say kept it unused
 > >> around.
 >
 > It also had a _very_ silly name! (which is better than having no-name
 > at all...).

well, Don Knuth was immediately able to place the quotation, ie where it was
derived from (see Prague interview, perhaps?), anyway, you are not seriously
claiming that it didn't make it because LaTeX is such a beautiful name, do you?

 > >> btw, i'm not even sure that we did ask that particular one, but we
 > >> certainly gave our input to features that are currently missing and
 > >> for which we had to produce horrible hacks to solve something or where
 > >> the current implementation would not always work in all situation
 > >> using normal TeX ...
 >
 > David Carlisle confirmed that 256 marks would be preferred to 16...

sure, i do prefer removing arbitrary restrictions as well, and yes i
welcome this mark enhancement even though i don't remember having
asked to get it extended from 16 to 256. In any way, we wouldn't have
asked something like that because we are currently working with e-tex
0.7 and found the limitation intolerable. what i was trying to point
out in my reply (to Uwe?) was that a lot of our input to e-tex came
from our experience with TeX in general.


 > Are there still features missing that you would like to see (other than
 > 16/32-bit/Unicode/OTPs/etc)?

yes:

  first of all, strip 16/32-bit/Unicode to request for a large enough unique
  internal addressing space (which is what is meant and needed)

  second, if you think feel otps are too radical then at least proper
  enhancements of multi-lingual interfaces, eg carrying lctables with
  the language nodes and what else is necessary in that area.

  third: to make your etc a little more concrete: a decent support page
  breaking

  forth: support for grid design, eg research and results into a different/additional concept
  of discrete glue that jumps rather than stretches

  fifths: ...

this is all in etex spirit right now meaning that this is what i would
like to see etex research go (and as far as the third goes where some
of my research interests lie too)

if that would be in etex and perhaps even if only the first two are
part of etex (as a proper solution to the third will need a lot of
time and might not fit into the etex paradigm) i guess i would put a
lot of weight into trying to get such a system accepted by a broad
basis.

but the same would be true if the omega project would incorporate etex
primitives and deals with the compatibility issues.

or if coming back to my suggestion we build a stable combined version
(probably also including pdftex near the end because that will have
matured by then too)


 > >> i know that this is a devil's loop that i'm describing, but i also
 > >> outlined how i see breaking that loop. if you think it could be broken
 > >> by writing some wonderful package that make people use e-tex then i
 > >> suggest you try that path.
 >
 > I don't think it could.  I _do_ think that if a future LaTeX were predicated on
 > e-TeX, and if that future LaTeX offered considerable advantages over the current
 > implementation, then that would of itself create a very rapid uptake of e-TeX.

true, except for: what are the considerable advantages we can currently and in the
foreseeable future offer? that's my whole argument and that is why...

 > You fear it would lead to the opposite -- a widespread refusal to move to the
 > new LaTeX -- and I respect your point of view; one day, perhaps we'll know
 > which of us is right...

i fear if staying with the status quo we will never find out, that is why (given
that i think it is about the right time) i would like to provide for a different
path of development or say an additional middle path.

 > >> but you would ask us to produce a LaTeX that would result in
 > >> non-usability for the majority of the users?
 >
 > With a sufficient overlap, this is not as bad as you make it seem.
 > If you were to release an e-TeX specific LaTeX on date D, but
 > continue supporting the older LaTeX until D' (D' >> D), this would
 > allow people adequate time to replace their ancient TeX 3.14s with
 > e-TeX 3.14159...  I would have thought an overlap of one year
 > would have been adequate, although I appreciate it would mean extra
 > work for the LaTeX tean; on the other hand, e-TeX is intended to
 > _reduce_ the load on the LaTeX team, by making many complex tasks
 > in TeX much simpler, so the load might balance out in the end...

one could have this at least initially as a separate project (by additional
volunteers outside the latex3 project --- my estimate would be a man month) and if
it turns out to be accepted have the project switch in at a later stage. right now
it means too much extra load, especially given that with with the same argument the
Omega project is leaning towards LaTeX development based on their system which
would be equally valid or not valid.

 > >> there are some areas in TeX that are simply wrong and where
 > >> compatibility is something for the historian. if you only accept
 > >> something as being a successor to TeX that not only keeps its good
 > >> parts but also its bad parts then that will be a poor successor.
 >
 > We _know_ that strict compatibility is essential; large organisations
 > such as AMS simply _will not_ replace TeX by anything, unless that
 > <anything> _guarantees_ that it will process all their legacy documents
 > in a manner identical to TeX.

okay we know that argument but then you can even have that, eg a compatibility
switch that produces that kind of compatibility you need --- just as

  \documentstyle{article}
  \author{frank\\
          \verb=frank{foo}@bar=}

runs without complaints although it loses the braces because that is
the way LaTeX209 typeset that article.

what i mean is that compatibility by turning off features or by restricting tables
to 256 rather than 16bit if asked for is not really an issue. and even if you
change the behaviour of a primitive you could turn that back in compat mode.

 > >> yes omega does not have a compatibility switch that can be set so that
 > >> it does behave like the original TeX but it does solve a large amount
 > >> of problems with TeX although only of a certain type. and what is the
 > >> problem with that? omega is more of an NTS system as e-tex is (which
 > >> did have different design goals) and there at least during initial
 > >> development compatibility is of no importance.
 >
 > See above...

see above what? see that i asked you what explicit compatibility problem you see
for a user or the ams using Omega. i claim that especially the ams wouldn't notice
any difference when running their production system through omega as it is
now. what argument did i miss?

 > I think that's all the relevant messages covered; the answers took longer

i fear i haven't covered half of the other important messages that came in the
recent days and i hope that people don't feel neglected when i have concentrated on
a few. and it took me a long time too although knowing me i'm not surprised :-)

 > than I intended, so I hope the LaTeX list adminstrators will forgive this
 > digression.  Please let me know _via private e-mail_ whether those who
 > would like to contribute to discussions on e-TeX would prefer to do so

to me this is not a discussion on etex but of much more general nature.  as long as
this is directly relevant to the future of LaTeX i don't mind staying with this
list and i didn't found that a digression at all; i personally fond it quite
helpful and i hope that others found this too.

good night
frank

ps this mail is at the limit of what latex-l let's you post so cut it down if you reply :-)
------- end -------

ATOM RSS1 RSS2