Received: by nummer-3.proteosys id <01C19443.A2F2743C@nummer-3.proteosys>; Thu, 3 Jan 2002 11:44:36 +0100 Return-Path: MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C19443.A2F2743C" x-vm-v5-data: ([nil nil nil nil nil nil nil nil nil][nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil]) X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message Subject: {2} Re: {1} Transition from LaTeX 2.09 to LaTeX 3.0 Date: Fri, 20 Mar 1992 14:07:59 +0100 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: From: "Frank Mittelbach" Sender: "LaTeX-L Mailing list" To: "Multiple recipients of" Reply-To: "LaTeX-L Mailing list" Status: R X-Status: X-Keywords: X-UID: 630 This is a multi-part message in MIME format. ------_=_NextPart_001_01C19443.A2F2743C Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subj: {1} Transition from LaTeX 2.09 to LaTeX 3.0 David said: > Various people have various bright ideas about the form that LaTeX 3 = might > take. Unfortunately, the project is being done on a voluntary basis, = with > hardly any funding, and with finite personnel. > > Is there anything to be said for not worrying too much about providing > a compatibility module for LaTeX 2.09? It would be one less thing to = do. I think a few words are in order. What kind of compatibility should be there? We have to distinguish between to areas: 1) User documents, i.e. syntax for describing the logical (and formatting) structure of documents 2) Style sheets and their internal coding. For 2) I would say that compatibility is out of question. If we like to achieve the goal of designing and implementing a suitable designer interface this means that we hopefully get rid of all internal kludges that are currently necessary to implement a certain layout in a style file. As a result the generation of a document-style should become much more straight forward, and therefore converting an existing style to the new syntax should be an easy task even if it is necessary to do this manually. For 1) the situation is slightly different. There will be certainly some changes in the command syntax to make the user interface more consistent and extend the functionality in certain areas. Mixing old and new syntax is probably something that is difficult to provide. Writing a ``compatibility style'' that implements the old syntax within the new program is in principle possible but a hard task too. On the other hand it is certainly necessary for people to run the thousands of already tagged documents when ltx3 is available. Now for this several strategies can be used: 1) run latex209 and ltx3 in parallel as David suggest 2) provide a compatibility style to run 209 documents within ltx3 3) provide a converter program that converts 209 sources to ltx3 sources. I'm sure that we will have 209 and ltx3 in parallel for some time. However, reproducing already finished documents is not the only possible application for documents written in 209 syntax. Much more important in my eyes is the ability to reuse documents are part of documents and at the same time make use of the additional functionality of ltx3. This is certainly not possible with a scheme like 1). Such a parallel running will only result that people use the frozen 209 and building stuff around it instead of switching over to the new system. As a result we will have all kind of confusion where people are searching for this and that for 209 requesting bug fixes etc. Even if one firmly says, no support on 209 any longer this means a lot of time-consuming work. And it certainly not helpful for ltx3. 2) is also not the right choice for similar reasons. First of all, writing such a beast will be non-trivial. Second you will soon find all kind of mixtures where people write partly in 209 and partly in ltx3 syntax and surely such mixtures are dangerous as it is nearly impossible to catch all possible kind of interactions between different concepts when applied interchangeably. Think of the problems that people currently have if the try to mix plain and latex commands without having a deep understanding of the interactions within the latex modules. As a result such mixtures nearly always bomb sooner or later at unsuspected places. For these reasons I think such a compatibility style is out. This leaves us with 3). Now, certainly writing a parser that inputs a 209 document and outputs a ltx3 version of it is non-trivial too. But in my eyes working on something like this is spending the time in a much better way then trying to do in the same time a comp-style. Of course, such a converter will probably need user interventions once a while, either because it isn't able to handle certain structures correctly or because there is a complicated mixture of plain and ltx commands. But even if the conversion needs some user help, if it gets 98\% of the document over this would be okay. Remember this is a one-time operation, it is necessary only once per document. Now the next question is how to write such a converter and in what language? Well I don't care about the language of the first version but the final version that gets distributed together with ltx3 will be probably written in TeX. Yes TeX, as this is the only language that happens to be around on every installation that has LaTeX. This will certainly be a slow converter but this doesn't matter. > This scenario might give "the project team" one less thing to do than = the > "provide a LaTeX 2.09 compatibility module for LaTeX 3.0" scenario. = Might > it be better to spend the finite time that is available looking to the > future than trying to be compatible with the past? Yes, not trying to be compatible to the past but allowing the past to catch up soon. Dr. Willibald Kraml commented on David > > I agree with this. I would even go one step further: should the name = "LaTeX > 3.0" be kept at all, which might suggest to all users a degree of > compatibility which might not be there? > Why not `NewLaTeX 1.0' or `NLaTeX 1.0' > > (I hope this discussion has not been raised before - after all: what's = in > a name?!)> No the discussion hasn't beed raised at least on a list. ltx3 is just a working name which may or may not change when we are about to give everything out. At the moment I think it is fine enough as it indicates at least a major release change. And from Phil: > > >>> Is there anything to be said for not worrying too much about = providing > >>> a compatibility module for LaTeX 2.09? It would be one less thing = to do." > > I would support this. There are sufficient kludges in LaTeX-2 that = the hassle > of emulating them surely outweighs the advantages. Let us take = advantage of > the total rewrite to \stress {get it right this time}. Yes, get it right this time, that's what I hope we can achieve, even with the limited resources. But even if there are sufficient kludges in LaTeX-2 there is one thing which is very important about latex209 and should be also there for ltx3. latex209 was stable for about 6 1/2 years this is most certainly one of the reasons for its popularity. ltx3 should and will come with a similar guarantee that code written for it and documents written in it will be processable at any ltx3 installation for at least a similar or longer period of time. So it is not only important to ``get it right'' but to get it right with the first try. Rolf Lindgren said: > I would support not caring too much about providing campatilbility. > Wouldn't it be more relevant to find out which options are most = popular, > and try to make them unnecessary or easy to reimplement? > > That way, most latex docs would port OK, which certainly should be a = goal. Yes certainly, I already said this in many messages and on talks. ltx3 will come with a broader range of standard styles and it will come in a way that converting the local variant of ``report'' etc. should be easy. I also requested to get send special document-styles so that we can what are the popular changes, etc. Finally during the beta test phase (or even earlier) I think of bowsing through all archives and converting the ``most important styles'' into new syntax. This would then also include options that provide coding. > BTW, does there exist some formal plan regarding what to support in = ltx3? > Is somebody compiling this? Can a draft be ftp'd from somewhere? This question is a bit too broad for me to answer. Yes there are plans and they are partly laid down in working papers. But so far there is nothing which can be ftp'd or something as none of these papers are in a shape to give them out, except perhaps for one. I will check and let you know on the list. Okay? Given your suggestion of finding out what options are popular. This is certainly a good idea. Only point is I would need a volounteer for this. Would you do this? I could think of something like a short info that you would like to get the following information, e.g. - what kind of latex style options are in use at the particluar installation - are there local variants? - if so, what are the originals? - what are the five or ten most popular document styles/ option-styles in use at the installation Perhaps some other questions. Answers then back to you so that you can compile a summary. A good starting point would be this list and I think everyone on the list should take five minutes to answer such questions (only don't send them over the list back :-) Would you take it, Rolf? (you see it is dangerous to open ones mouth) Frank ------_=_NextPart_001_01C19443.A2F2743C Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable {2} Re: {1} Transition from LaTeX 2.09 to LaTeX 3.0

Subj:   {1} Transition from LaTeX 2.09 to = LaTeX 3.0


David said:

> Various people have various bright ideas about = the form that LaTeX 3 might
> take.  Unfortunately, the project is being = done on a voluntary basis, with
> hardly any funding, and with finite = personnel.
>
> Is there anything to be said for not worrying = too much about providing
> a compatibility module for LaTeX 2.09?  It = would be one less thing to do.

I think a few words are in order. What kind of = compatibility should be
there? We have to distinguish between to = areas:

 1) User documents, i.e. syntax for describing = the logical (and
formatting) structure of documents

 2) Style sheets and their internal = coding.

For 2) I would say that compatibility is out of = question. If we like
to achieve the goal of designing and implementing a = suitable designer
interface this means that we hopefully get rid of all = internal kludges
that are currently necessary to implement a certain = layout in a style
file. As a result the generation of a document-style = should become
much more straight forward, and therefore converting = an existing style
to the new syntax should be an easy task even if it = is necessary to
do this manually.

For 1) the situation is slightly different. There will = be certainly
some changes in the command syntax to make the user = interface more
consistent and extend the functionality in certain = areas. Mixing old
and new syntax is probably something that is = difficult to provide.
Writing a ``compatibility style'' that implements the = old syntax
within the new program is in principle possible but a = hard task too.
On the other hand it is certainly necessary for = people to run the
thousands of already tagged documents when ltx3 is = available.

Now for this several strategies can be used:

 1) run latex209 and ltx3 in parallel as David = suggest

 2) provide a compatibility style to run 209 = documents within ltx3

 3) provide a converter program that converts 209 = sources to ltx3
sources.

I'm sure that we will have 209 and ltx3 in parallel = for some time.
However, reproducing already finished documents is = not the only
possible application for documents written in 209 = syntax. Much more
important in my eyes is the ability to reuse = documents are part of
documents and at the same time make use of the = additional
functionality of ltx3. This is certainly not possible = with a scheme
like 1). Such a parallel running will only result = that people use the
frozen 209 and building stuff around it instead of = switching over to
the new system. As a result we will have all kind of = confusion where
people are searching for this and that for 209 = requesting bug fixes
etc. Even if one firmly says, no support on 209 any = longer this means
a lot of time-consuming work. And it certainly not = helpful for ltx3.

2) is also not the right choice for similar reasons. = First of all,
writing such a beast will be non-trivial. Second you = will soon find
all kind of mixtures where people write partly in 209 = and partly in
ltx3 syntax and surely such mixtures are dangerous as = it is nearly
impossible to catch all possible kind of interactions = between
different concepts when applied interchangeably. = Think of the problems
that people currently have if the try to mix plain = and latex commands
without having a deep understanding of the = interactions within the
latex modules. As a result such mixtures nearly = always bomb sooner or
later at unsuspected places. For these reasons I = think such a
compatibility style is out.

This leaves us with 3). Now, certainly writing a = parser that inputs a
209 document and outputs a ltx3 version of it is = non-trivial too. But
in my eyes working on something like this is spending = the time in a
much better way then trying to do in the same time a = comp-style. Of
course, such a converter will probably need user = interventions once a
while, either because it isn't able to handle certain = structures
correctly or because there is a complicated mixture = of plain and ltx
commands. But even if the conversion needs some user = help, if it gets
98\% of the document over this would be okay. = Remember this is a
one-time operation, it is necessary only once per = document. Now the
next question is how to write such a converter and in = what language?
Well I don't care about the language of the first = version but the
final version that gets distributed together with = ltx3 will be
probably written in TeX. Yes TeX, as this is the only = language that
happens to be around on every installation that has = LaTeX. This will
certainly be a slow converter but this doesn't = matter.

> This scenario might give "the project = team" one less thing to do than the
> "provide a LaTeX 2.09 compatibility module = for LaTeX 3.0" scenario.  Might
> it be better to spend the finite time that is = available looking to the
> future than trying to be compatible with the = past?

Yes, not trying to be compatible to the past but = allowing the past to
catch up soon.

Dr. Willibald Kraml commented on David
>
> I agree with this. I would even go one step = further: should the name "LaTeX
> 3.0" be kept at all, which might suggest to = all users a degree of
> compatibility which might not be there?
> Why not `NewLaTeX 1.0' or `NLaTeX 1.0'
>
> (I hope this discussion has not been raised = before - after all: what's in
> a name?!)>

No the discussion hasn't beed raised at least on a = list. ltx3 is just
a working name which may or may not change when we = are about to give
everything out. At the moment I think it is fine = enough as it
indicates at least a major release change.


And from Phil:
>
> >>> Is there anything to be said for = not worrying too much about providing
> >>> a compatibility module for LaTeX = 2.09?  It would be one less thing to do."
>
> I would support this.  There are sufficient = kludges in LaTeX-2 that the hassle
> of emulating them surely outweighs the = advantages.  Let us take advantage of
> the total rewrite to \stress {get it right this = time}.

Yes, get it right this time, that's what I hope we can = achieve, even
with the limited resources.

But even if there are sufficient kludges in LaTeX-2 = there is one thing
which is very important about latex209 and should be = also there for
ltx3. latex209 was stable for about 6 1/2 years this = is most
certainly one of the reasons for its popularity. ltx3 = should and will
come with a similar guarantee that code written for = it and documents
written in it will be processable at any ltx3 = installation for at
least a similar or longer period of time. So it is = not only important
to ``get it right'' but to get it right with the = first try.


 Rolf Lindgren said:

> I would support not caring too much about = providing campatilbility.
> Wouldn't it be more relevant to find out which = options are most popular,
> and try to make them unnecessary or easy to = reimplement?
>
> That way, most latex docs would port OK, which = certainly should be a goal.

Yes certainly, I already said this in many messages = and on talks. ltx3
will come with a broader range of standard styles and = it will come in
a way that converting the local variant of ``report'' = etc. should be
easy. I also requested to get send special = document-styles so that we
can what are the popular changes, etc.

Finally during the beta test phase (or even earlier) I = think of
bowsing through all archives and converting the = ``most important
styles'' into new syntax. This would then also = include options that
provide coding.

> BTW, does there exist some formal plan regarding = what to support in ltx3?
> Is somebody compiling this? Can a draft be ftp'd = from somewhere?

This question is a bit too broad for me to answer. Yes = there are plans
and they are partly laid down in working papers. But = so far there is
nothing which can be ftp'd or something as none of = these papers are in
a shape to give them out, except perhaps for one. I = will check and let
you know on the list. Okay?

Given your suggestion of finding out what options are = popular. This is
certainly a good idea. Only point is I would need a = volounteer for
this. Would you do this? I could think of something = like a short info
that you would like to get the following information, = e.g.

- what kind of latex style options are in use at the = particluar
installation

- are there local variants?

- if so, what are the originals?

- what are the five or ten most popular document = styles/ option-styles
in use at the installation

Perhaps some other questions. Answers then back to you = so that you can
compile a summary.  A good starting point would = be this list and I
think everyone on the list should take five minutes = to answer such
questions (only don't send them over the list back = :-)

Would you take it, Rolf? (you see it is dangerous to = open ones mouth)






Frank

------_=_NextPart_001_01C19443.A2F2743C--