%Copyright: C.G. van der Laan, cgl@rc.service.rug.nl, 31 5941 1525
%Version Aug, 1994
\input blue.fmt           %\tolerance500\hbadness=499\hfuzz=4pt
\everyverbatim{\emc}
\postverbatim{\smallbreak\noindent}
\let\titlefont\bf
\title{BLUe's Format}
\subtitle{---The best of both worlds\footnote*{%
   blue.fmt is released at Euro\TeX'94. How it will develop depends
   much on how it will be received.
   It is not perfect nor complete for a general audience,
   but serves my purposes.
   BLUe stands for Mr BLUe---my innocent user
   and relative of Ben Lee User of the
   \TeX book fame.}}
\issue{CTAN, NTG's fileserver, ***PRERELEASE***}
\keywords{
Active documents, design, documentation,
education, error handling, floats, format,
inner versus outer world, inserts,
lifephases formats,
macro writing, markup language, plain \TeX,
scripts, software engineering,
(reusable) software parts,
style, transparencies.
}%\endkeywords
\abstract{An independent format---blue.fmt---is provided to assist
authors with creating, formatting, exchanging and maintaining
compuscripts during the lifephases of publications.
The format  builds upon manmac.tex and
the functionalities provided by tugboat.sty.
Experience gained by publishers has been picked up too,
because of my in-depth study of the activities of AMS
with respect to \TeX{} formatting.
More recent  work of Knuth and co-authors has been borrowed from
gkpmac.tex.
The design goal was to provide a format
which suits me,
which is easy to customize---to the world
outside, and in general to changing circumstances---and
which adheres to the adages of software engi\-neer\-ing.
Another aim of blue.fmt is that it can be used throughout the
life-cycle of publications on modest equipment, with a wink to
active documents, and that it will serve a lifetime.
En-passant the design process is accounted for.
New is the handling of a database of
references with cross-referencing, all in one-pass job.
Even newer is the proposal for a database of pictures,
from which specified pictures can be loaded selectively.
}%end abstract
\thisreferences{The \TeX book and Lamport's \LaTeX{} user's guide
   are omni-\-pre\-sent and not explicitly listed.
  \hfuzz=7pt}%\lsstraight
\references{\beebenhfd
 \berryka
 \cheswickb
 \doobma
 \jonesdm
 \grahamrla
 \greenwadegd
 \hamiltonh
 \heckelp
 \hendricksonaa\hendricksonax
\laancgc\laancge\laancgg\laancgga\laancgh\laancgi\laancgj\laancgk
    \laancgl\laancgm\laancgp\laancgq\laancgr\laancgs\laancgt
    \laancgu\laancgw\laancgx\laancgy\laancgz
 \salomondd\salomondg
 \spivakmdb
 \taylorpc
 \tugpub
 \whitneyrf
}%end references
\pictures{\bluemodelpic\furutapic\ntgpiepic\regulartransparencypic
          \vectorbundlepic}
\script{%\beginscript for test purpose the minimal form
\head{Introduction}
After having studied manmac.tex---Knuth's macros for formatting
his Computers and Typesetting series---as well as the
TUGboat styles---the (plain) format for TUGboat, the bulletin
of the \TeX{} Users Group---I decided to combine the best of
both worlds.
The goodies given by Knuth to the world
are merged with the achievements offered by the
members of the various \TeX{} users groups.
In order to do this gracefully, I first had to simplify
tugboat.sty.\ftn{See my TUGboat BLUes for the proposal tug.ppt,
   which will be called ppt.tex from now on.}
This superset is extended by macros which have emerged
from later work of Knuth,
and by some macros of my own,
dictated by the functionalities I need.
Experience gained by publishers  has influenced this work too
because of my in-depth study of the activities of AMS
with \AllTeX{} as formatting tool.\ftn{Because AMS has played a leading
   role in the \TeX{} world and because TUGboat's chief editor
   is employed by AMS, it is not surprising that much of AMS'
   experience has already been incorporated in tugboat.sty.
   For an account of the activities of AMS with \TeX{} as
   formatting tool see `AMS BLUes' and the references given in there.}
I decided to separate outer level markup from inner level markup.
The idea behind this separation is to identify inner level macros
which are  context independent---or nearly so---and
which can be used as a toolbox along with |<Any>|\TeX.\ftn{Up
   till today this borderline between inner and outer level
   is ill-defined. Definitely (overlapping) clusters can be
   identified.}
By splitting off as much as possible to the inner level
the outer level will shrink, with the pleasant side-effect
that the amount of work to maintain the format will diminish.
To adapt a script for other environments---and changing
circumstances in general---will be simpler too.
How to cope with the outside world of computer-assisted typography
in relation with your inside world, lies at the heart of this article.
\beginquote
The world outside to learn from, to assimilate what is there,
and to give back what you have lived through, to close the circle.
\endquote
\subsubhead{Warning.} In this composition I made use of earlier
work of myself
and adapted among others the more systematic use of token variables
instead of defs for |\pre<tag>| and |\post<tag>| and the like
next to the systematic
building of  one-part macros on top of two-part macros.
Be aware of these anachronisms in the older articles.
\subhead{Why?} In a sense this paper is a summing up of my research
of what others have accomplished before in the area of
computer-assisted typography via \TeX.
I strove after introducing as little overhead as possible---the
paper and the macros should not be unduly long.
I respect my readers and don't like them to waste their precious
time, so I worked hard to make reading  worthwhile.
\beginquote
There is always something to  learn from macros
written with care.
\endquote
The design goal of blue.fmt was to create a format
which obeys the software engineering adages
consistency simplicity, intelligibility, portability, flexibility,
extensibility, stability, and openness, next to adaptability,
pleasant documentation and correctness of course :-).
Furthermore, I don't like superseding packages, which neglect
other activities.
For example when  other activities can't be incorporated easily,
or even worse when they are forbidden.
I favour a toolbox-like  approach,
which builds upon plain and allows access to lower layers,
let us say I feel myself happy with open systems.
The file is considered a format because that loads fast.\ftn{Because of
   my default value for \cs{footline} a page will be set when
   creating the format. I kludged the page counter into
   -1 at the end of the format to counteract the increase.}
\smallskip
The short answer to the `Why?' comes down to
\beginquote    I need a format\ftn{Read: order all the stuff.}
   which assists computer-assisted handling of information,
   be it consumption or production and dessimination, and
   which serves a lifetime.
\endquote
Another answer to the Why? is that when presenting Manmac's BLUes at GUST 94
I was asked to provide a user's guide for manmac. BLUe's Format is my
answer to that request, although this article is not a user's guide, per se.
The audience in mind consists of those authors who
\bitem practise (self-)publishing
\bitem choose for English and ASCII
\bitem adhere to \TeX{} formatting
\bitem like a lifetime tool, sta\-bility,
       consis\-ten\-cy, simpli\-ci\-ty, porta\-bi\-li\-ty,
       gene\-ra\-li\-ty, flexi\-bi\-li\-ty
\bitem favour an open, well-documented system
\bitem prefer an extensible (formatting) language
\bitem and support the PD software adage.
\noindent Pondering about the audience a little longer makes me
feel that most of them belong to the {\it unorganized\/}
\TeX{} users, who happily use the tool without much ado.
However, the  style-designer and hacker can find
something new in here too,
especially the handling of bibliographies in a one-pass job,
respectively  how to code a minimal markup macro on top of
a two-part macro with the same functionalities.
Whatever the value of blue.fmt it definitely servers an educational
purpose, with a wink to active documents.\ftn{I like teaching a
   Manmac BLUes tutorial where the
   What? and How-to? are dealt with, prolongated
   into a workshop with goodies from blue.fmt
   and those contributed by the participants.
   A really advanced class about macro writing.}
It also shows how to build a collection of macros from
various sources, or to say it straight to keep things in order.
\subhead{Notations and definitions.}
AMS denotes the \AMS.
manmac.tex stands for the macros used by Knuth for formatting
his Computers and Typsetting series of books.
ppt.tex stands for my macros inspired by tugboat.sty,
to obtain a similar layout.
gkpmac.tex stand for the macros used by Knuth and his co-authors
to format `Concrete Mathematics.'
The picture subset taken from this collection is called
gkppic.tex.
TUG stands for the international \TeX{} Users Group.
TUGboat is TUG's bulletin. TTN stands for TUG's newsletter:
\TeX{} and TUG News.
NTG stands for the Dutch Language-oriented users group.
MAPS is NTG's bulletin.
AN is the abbreviation for automatic numbering.
The name BLUe has its history, the filename is in lowercase.
Control sequence names consist of letters only and therefore
when a hyphen occurs in this text
as part of a name, it is because of the hyphenation.
I like to denote the marked up file as a script---a compuscript,
if you like---independent from whether the result is a publication,
a set of transparencies, or you name it.
A document is a publication on paper, be it in the
form of a report, an article or a book.
A multi-media publication is a script
which can be processed into forms to be viewed
via several media.
Names of the control sequences consist of a root part
with prefixes and postfixes.
\cs{ag} and \cs{ea} are shortcuts for \cs{aftergroup}
respectively \cs{expandafter}.
\head{Specifications}
In designing a format there is not such a thing as {\it one\/}
specification. We have to deal with the specifications for
\bitem look-and-feel in print---typographical design
\bitem markup language---syntax and semantics
\bitem coding style---data types, control sequences.
\noindent Pondering about this and being aware of the
difficulties the software engineering history has made explicit,
it is not much of a surprise that there is
no format that has been designed starting from
the specifications.\ftn{Except for manmac, I presume.}
Even not tugboat.sty, which was `designed' on the fly and
pushed forward by the need of advanced \TeX ies.
This can be understood because format design is a new activity,
and software engineering has been proven to be a very difficult
area.
Moreover, the work has to be done mainly by volunteers.
\head{Typographical design}
Being a layman on the issue I considered it best to take over what has
been in use for a dozen of years.
For the look-and-feel of the result of a blue script in two-column
I took over the main aspects of the TUGboat layout.
\head{Markup language}
When pondering about developing a format,
one has to realise that essentially
a markup language will be created.
That means that the designer has to be explicit about
the syntax and the semantics, and above all provide
a simple user interface.
That is really something in the presence of
macro expansion, because everything can be changed on the fly.
I like the orthogonal approach, for example to treat
\beginquote
\xyz{Minimal tags}{Options}{Place within context}
\endquote
independent from each other. This approach has been on my mind
throughout blue.fmt, but not applied unduly.
\subhead{Conventions control sequences.}
I chose for
\bitem |\begin<tag>|, |\end<tag>| pairs, which take global options via
   |\every<tag>| and local options via |\this<tag>|\ftn{In analogy
     with Knuth's \cs{every}$\langle tag\rangle\{\dots\}$.}
\bitem |\<tag>|, minimal variant
\bitem token variables  |\this!-<tag>| and |\every!-<tag>|
      for user guidance for the inner markup
\bitem token variables |\pre!-<tag>| and |\post!-<tag>|
    for controlling positioning within context
\bitem token variables  to convey information, like the title
\bitem |\pasteup<tag>|-s, which position the formatted script
   element within context.
\noindent In a nutshell this comes down to the following control
sequences\ftn{Because blue.fmt is a composite these
   conventions hold mainly for the outer markup. The inner markup
   control sequences can originate from other sources with
   their own conventions about naming.}
\beginverbatim
\<tag>{...}     or     \begin<tag>
                        ...
                       \end<tag>
!endverbatim%the next indented by \parindent, looks better
optionally preceded by
\beginverbatim
\every<tag>{...}   and/or   \this<tag>{...}
!endverbatim
When the place within context should differ from
the default supply\ftn{This does not hold for the title part
   elements. Too much nitty-gritty. Not only the positioning within
   context is relevant but also the ordering.}
\beginverbatim
\pre<tag>{...}   and/or   \post<tag>{...}
!endverbatim
I did not bother about the size of the tags, because with
nowadays \AllTeX{} intelligent editors these tags don't have to be
typed in anymore.
Moreover, for \TeX{} each tag represents one token
independent of its length.
\subsubhead{The bad news} is that
minimal markup for \cs{head} and the like can be
felt as an anachronism.
With \AllTeX{} intelligent editors we don't need minimal markup
anymore. The reason to process headtitles and the like on the
fly is that catcodes are assigned in phase while processing,
 meaning for example we can use verbatims as part of the title.
It is not possible to associate, via my \TeX nique explained
later with \cs{head}, a one-part
macro to every two-part macro, in full generality.
\subsubhead{The good news} is that via this mechanism we can abstract
to a high degree at the {\em user level\/} from token variables
and definitions. That is
the user can supply the information as if it was a token variable.
\subsubhead{Options.} For scripts a useful application is
to specify whether the last script is at hand or not.
Another use is to specify \cs{onecol}.
With verbatims it is handy to supply catcode changes
and to specify a file for verbatim file input.
The attributes for tables can be specified via this mechanism,
for example \cs{ruled}, \cs{framed} and the like.
Pictures can take different values from the default for their size.
\subsubhead{Place within context.}
This is a very general issue. In fact formatting can be seen as
placing elements within context. So why bother?
To parameterize the place within context is a good approach
because it provides for a user, or style designer, the hooks
needed for the look-and-feel of the result.
A good example of this is \cs{head} and the like. A user can determine
the whitespace via
\beginverbatim
\prehead{...}       and       \posthead{...}
!endverbatim
Ditto for verbatims.
For tables the \cs{postbtable} token variable is already used for
measurements of the table. Usually I place the vbox,
which results from my btable macro, in a math
display, conform the examples in the \TeX book.
For systematic placement of tables either mechanism can be used.
I'm still pondering about  to format tables at the
beginning and past up these boxes  at the right place.
The difference with references is that we have usually
many tables, and box registers are limited available.
The next best idea is to have tables also stored in a database
and load from there selectively, similar to what I  already do with figures.
For the moment there is nothing special about figures to be included
in the database for selection, as long as it is a
definition. So add your tables as well,
and use them similarly.
For figures I did not provide a \cs{prefigure} or \cs{postfigure}
of yet. The pictures are delivered in a hbox.
\subsubhead{Changing the defaults} in token variables
can be done in two ways
\bitem assign new values to the token variables
\bitem extend the token variable with your extras.
The latter can be done as follows for the cases of appending
to  |\post<tag>| and preceding to |\pre<tag>|.\ftn{The latter is inspired
    by Salomon's courseware.}
\beginverbatim
\post<tag>\ea{\the\post<tag> <your extensions>}
\def\preadd{<your extensions>}
\pre<tag>\ea{\ea\preadd\the\pre<tag>}
!endverbatim
\subsubhead{Processing more than one} script per job must be possible.
For example for a journal when several articles,
each as a selfcontained script,  have to be formatted in total,
with similar headers and footers.
\subhead{What to start with?}
The script should start with what I call {\em information\/} elements,
preliminary matter such as title, keywords and the like,
next to supplying references and pictures.
All in arbitrary order and done before \cs{beginscript}.
Token variables have defaults. For example your blue.fmt
has the token variables \cs{author}, \cs{address}, and
\cs{net\-address} already filled-in with defaults. If you wish to assign
values to these token variables at the start of your compuscript,
feel free to do so. That is the way of substituting
actual values for the defaults. Handy isn't it?
The contents proper should start with \cs{begin\-script},
              to handle the paste-up of the title matter,
and end with \cs{end\-script}, to finish the script.
This can be preceded by \cs{every\-script} or \cs{this\-script}.
When a script is
formatted on its own the use of \cs{this\-script}, and especially
\cs{every\-script} can be omitted.
For processing more than one script provide
\cs{not\-last\-script} in \cs{every\-script}, and
provide the last script with \cs{last\-script} in
\cs{this\-script}.
However, most of the above does {\em not\/} hinder that
\thisverbatim{\catcode`!=12
  \catcode`+=0 }
\beginverbatim
Thank you!\bye
+endverbatim
can be run with blue.fmt, although it is not a blue script according
to my definitions.
It is very handy for testing parts of a blue script.
\beginquote
blue.fmt does not impose!
\endquote
\subhead{Active documents.}
The revolution of computer-assisted typo\-graphy
has introduced the concept of active documents.
For me this activity is all about
\beginquote
the discrepancy of the ordering of elements as marked up
in the compuscript, and the positioning and representation
of the document elements in the resulting document.
\endquote
The hierarchy in which the document elements have been supplied
does no longer imply a similar sequence in print.
The ordering can be altered due to writing to and reading
from the computer's memory, be it  RAM (random access memory)
or be it external memory with sequential access.\ftn{Knuth was aware of
   this avant l\`a lettre via his insert mechanism and his
   coding of answers.}
Actually this happens with the title and the like.
The keywords, abstract and contents are set in boxes
and pasted up in the replacement text of
\cs{beginscript}.
Similarly the references have to be specified in the
preliminary part.
\beginquote
The place to paste up
the references can be chosen  by the
author via \cs{pasteupreferences}.
\endquote
Similar for pictures. These are also selectively loaded
in the preliminary part and
\beginquote
the place to paste up a picture can be chosen
by the author via the use of  |\<name>pic|.
\endquote
In order not to lose ground, I will start from an informal
template and formalise from there.
The result aims at being {\em supportive\/} for authors,
a structure to which marked up documents
should comply to.
%The meaning of the tags must be as provided in blue.fmt.
\subhead{Template.}
When I start writing down my findings, I involuntarily
use an article-like structure, which at the outer level bears
the  markup as can be seen from
the accompanying table.\ftn{After having studied various works
   it is hard  to get rid of anachronisms. Being aware of that
   I hope it did work out. For the code of the structures given in the
   template start with the ppt part of the blue.fmt file.}
In contrast with summing up all what is provided in manmac.tex, gkp.mac
and ppt.tex,\ftn{See for that my Manmac BLUes,
   BLUe's Graphs, respectively TUGboat BLUes.}
I will start informally
with the desired functionalities---at two levels---%
and indicate which macros will accomplish these.
The template---as a sketch of what I have in mind---should guide me
towards what is needed syntactically and semantically.
\subhead{Outer markup.}
Generally I use a template, an empty article as fill-in form,
with the main markup structures already there and in the
right order. A rudimentary discussion follows below, because
once used to the template, the outer markup is a trifle.
\topinsert
\thisverbatim{\catcode`\~=14 }
\beginverbatim
%Template script---outer level markup
%Info part: elements in arbitrary order
\issue{MAPS 94.2}
\title{BLUe's Cross-Referencing}
\subtitle{---A one-pass approach}
\keywords{...}
\abstract{...}
\contents{Introduction
  \quad Why?
  ...
  Acknowledgements, Conclusion, References}
\references{~
\!math!langle!,name_1!rangle!dots!endmath~
\!math!langle!,name_n!rangle!endmath}
\pictures{~
\!math!langle!,name_1!rangle!,pic!dots!endmath~
\!math!langle!,name_n!rangle!,pic!endmath}
\beginscript                 %Copy proper
\head{Introduction}
%various parts with inner markup
%Back matter (with simple index to come)
\pasteupreferences
\endscript
!endverbatim
\endinsert
\subsubhead{Information part} consists of title,\ftn{I favour a short
   title, optionally extended by a subtitle. Because of this
   I can use the title in the running head too, and there is no
   need for a separate specification of a running title.}
author (address information),
keywords, abstract, contents,  references, and pictures
independent from whether
these issues appear at the beginning in print or not.\ftn{In
   regular issues of TUGboat the address information is
   printed at the end, while in the proceedings issues the
   address information is printed as part of the front matter.}
The idea is to supply this information early in the script---and in
fact author and address information are suited for defaults---such that
it can be used anywhere.\ftn{This is reminiscent of providing it
   at the beginning and typesetting it immediately. Nowadays
   the providing of the information is no longer tied up with
   formatting it immediately.}
By supplying the references also at the beginning we don't need
a multi-pass job for cross-referencing.\ftn{The references are set
   and the names get their numbers as replacement text. Later in the
   script the markup command \cs{pasteupreferences} pastes up the
   earlier set references.}
Bibliographical information\Dash classification, ISBN and the like\Dash
is added separately and generally by other
tools.
It is relevant when the work enters the outside world.
So the publisher will take care of these issues, I guess.
The \cs{beginscript} control sequence  marks the beginning
of the compuscript proper.\ftn{Via this idea of scripts an editor
    can happily run
    \cs{onecol} and \cs{twocol} submissions in one job.}
Whether it concerns an article,  report,  book or  a set of
transparencies, \cs{beginscript} is a generic control sequence
which has been defined appropriately within the context.
It starts formatting the title part and processing the script elements
which follow.
The compuscript is ended by \cs{endscript}.\ftn{And if you are a masochist
   you can use the minimal from.}
The keywords and abstract parts traditionally serve the quick reader
and facilitate the selection process,
whether the article is about what you are looking for.
Note that the contents is context dependent and
presumes the table of `contents' to be supplied
in a {\em near natural\/} way: line-by line.
For indentation\ftn{Not for
   transparencies.} simply supply
  \cs{quad} or \cs{qquad}.\ftn{No automatisms have been provided
   for automatically collecting the table of contents information
   while formatting. No multi-pass job!}
\subsubhead{Chapters and the like} are marked up with rootname
|head|, with the title as argument.\ftn{Similarly for |subhead|,
   and |subsubhead|.}
I hardly use automatic numbering, and when needed it can
be added easily.
Note that I like running-in heads---like in this paragraph---and therefore
no enddot is inserted automatically.\ftn{Of course I could have
   parameterized this in \cs{end\-punc\-tu\-ation}.
   For the moment I refrained from doing so.
   Knowing when to stop is important.}
\subsubhead{Running headers and footers}
   make use of the token variables \cs{headline} and \cs{footline},
   with the information automatically distilled from the title
   token variables, like \cs{title}, \cs{author}, and \cs{issue}.
\subsubhead{Bibliography} handling is a much overstressed aspect IMHO.
I like to keep it simple. Firts of all I like that the database
of formatted entries is already with the publisher, and that we only
have to refer to the entries. But that seems wishful thinking still,
although my approach comes close.
I use the macros and database which emerged from `BLUe's Bibliography,'
because it facilitates the selection from the database---the file
lit.dat---and performs the formatting all in one-pass job.
The typograhical design  has been inspired by  ams.ppt style.
It is  customizable.
Since `BLUe's Bibliography' has been dis\-tri\-bu\-ted in MAPS,
I extended the macros with a simpler user interface.
\beginquote
After \cs{be\-gin\-re\-fer\-en\-ces} just supply
the names for the references and close by
\cs{end\-re\-fer\-en\-ces}.\ftn{Or use the minimal markup variant.}
And paste up via \cs{pasteupreferences}.
\endquote
Text to precede the list of references can be supplied
via \cs{this\-ref\-er\-en\-ces}, as has been done in this article.
The above can best be provided in the preliminary part, definitely
before a reference to a literature item by number will be made.
The paste-up is a different matter, and the place is at the
discretion of the author.
Cross-referencing  can be
done via the name-and-year convention\ftn{Fundamental for me are
   author names and the year of publication,
   next to the titles and sources.
   In the field I'm working I know by heart which
   work is hinted at when the author and year is mentioned.}
or via the \TeX nique as explained in `BLUe's Bibliography,'
that is each name will be redefined by its sequence number
as replacement text at the place where the references are supplied.
Furthermore, I'm so much surprised by the ample use of punctuation
in items of a literature database. I favour minimal punctuation.
Abstracting from this punctuation was another option, but I decided
not to try for a too general approach with its overhead.
\beginquote
Minimalism as a royal road to simplicity!
\endquote
I agree that my references macros do a lot of work.
It looks a bit overdone when only a few references have to be formatted.
Firstly, so what? The references have already been used
before, so they don't need proofreading so much,\ftn{Nor the
   fine-tuning with respect to underfull box messages, not to speak
   about avoiding overfull ones. See Phil Taylor's paper
   on the issue.}
and therefore can be suppressed while working on the paper.
Secondly, if you don't like BLUe's automatisms
why not format a list of the few references yourself?
Just a heading and some \cs{items}, that is all. See the
\TeX book p.340--341, for the example article of Knuth.\ftn{If you prefer
   to mark up all the subelements of an bibliography feel free to
   do so. For formatting via blue.fmt \cs{ls} must be customized then.
   Available are \cs{lsams}, default, and \cs{lsstraight}.}
\subsubhead{Pictures.}
The use is to supply the following in the
preliminary part
\thisverbatim{\catcode`~=14 }
\beginverbatim
\pictures{~
\!math!langle!,name_1!rangle!,pic!dots!endmath~
\!math!langle!,name_n!rangle!,pic!endmath}
!endverbatim
\noindent and to invoke the picture at the appropriate place
in the script via
\beginverbatim
\!math!langle!,name!rangle!,pic!endmath
!endverbatim
\noindent optionally preceded with |\thispicture{...}| .
\subsubhead{Index} preparation comes usually later.
It can be done via
manmac's writing  of index reminders (IRs) to a file,
and formatting the sorted, compressed and in general
enriched IRs via manmac's doublecolumns environment.
The IRs written to the file index.tex obey the syntax
\beginverbatim
<word>!]!!<digit>!]<page number>
!endverbatim
Knuth sorts and enriches the IRs outside of \TeX.
The resulting marked up file contains elements like
\thisverbatim{\catcode`\|=12 }
\beginverbatim
*|\accent| (general accent), 9, 54, 86, 283,
   +286.
accents (\'{} \`{} \"{} etc\period), 7,
   @52--53, @339, +356, 357, 427--429.
\sub as ligatures, 46, 54.
\sub in math, @135--137, 141, 164--165, +359,
   435, +443.
\sub on top of accents, 136.
\sub table, 52, 135, 339.
!endverbatim\noindent
Note that he has added the markup
\begindisplay\displayindent\parindent
*        &primitives of \TeX\cr
@        &set in italics\ftn{An instructive example.}\cr
+        &underline\ftn{ Definition or main source of information.}\cr
\cs{sub} &a subentry
\enddisplay
next to `see also' text and other helpful goodies.
In my `Sorting in BLUe' the sorting within \TeX{} of IRs is
touched upon. I agree with Knuth that an index enriched in his way
can better be handled by several jobs, to obey the separation of
concerns principle.\ftn{\TeX{} to provide for the index.tex file
   $\rightarrow$ a bunch of (Perl) scripts to sort and markup the
   information
   $\rightarrow$ \TeX{} again to format the index.}
\goodbreak However,
\beginquote
it would be nice to be able to process a simple index
completely and automatically within \TeX.\ftn{See `BLUe's Index.'}
\endquote
Makeindex comes with \LaTeX, and Salomon has provided a plain version
of it, to work together with some Modula2 programs.
These programs did not adopt Knuth's markup for the IRs.
I guess this is because they also provided for tags to mark up
subentries, see also and the like.
The macros provided in manmac for writing IRs to a file
and for formatting the enriched information in two columns are superb.
\subhead{Inner markup.}
In contrast with the outer markup I will include in this
part examples borrowed from the special papers
on the issue.\ftn{For more details peruse my BLUe's series
   as made available  in NTG's MAPS, or await the upcoming book
   'Publishing with \TeX.' The latter also accounts for
   adjustments of details, the mods,
   the increased consistency, the ease of use and the like.
   Once I overheard George Greenwade---the originator of CTAN---%
   saying that he was so much
   surprised by the number of changes for submissions to the CTANs.
   And indeed I must confess I changed quite some details too.}
At the end I will come back to manmac's one-column specifics.
I won't treat the abbreviations and logos taken from tugboat.cmn.
These are self-explanatory.
\subsubhead{(Special) Paragraphs}
are for example provided by \cs{item}, and derivatives.
\example{Bulleted items}
I love Knuth's \cs{item}. It is so general and can easily be adapted
to automatically format bulleted items---and others as well---via the
 use of\ftn{\TB, exercise 14.10.}
\beginverbatim
\def\bitem{\item{$\bullet$}}
!endverbatim
\example{AN items}
Confer \TB, p102
\beginquote
`It is customary to put \cs{medskip} before and after a group of
 itemized paragraphs, and to say \cs{noindent} before any closing
 remarks that apply to all of the cases.'\ftn{I did not chose
   for \cs{medskip} nor did I impose the \cs{noindent}.
   I did not insert special markup preceding the list of items.}
\endquote
The markup reads
\beginverbatim
\nitem first\\second line
\itemitem{} subitems\\next line
\nitem second\\second line
\smallbreak
!endverbatim
\noindent with result
\smallskip
\nitem first\\second line
\itemitem{} subitems\\next line
\nitem second\\second line
\smallbreak
Other special paragraphs are formed by footnotes
endnotes and notes in the margin. For footnotes with automatic
numbering I use \cs{ftn}, based upon \TB{} exercise 15.12.
This exercise is an extension of \cs{footnote}
which processes the footnote on the fly, and therefore allows
verbatims.
\example{Footnote with AN\ftn{From TUGboat BLUes.}}
\beginverbatim
   ...text.\ftn{Note to text.}
!endverbatim
\subsubhead{Quotations} as used in the \TeX book are right justified.
The markup for the end of page 1 of the \TeX book reads as follows.
\example{\TB~p.~1}
\beginverbatim
%chapter contents precedes
\endchapter%Followed by the quotations
They do certainly give very strange and
new-fangled names to diseases.
\author ^{PLATO}, {\sl The Republic},
Book 3 (c.\ 375 B.C.) % 405c
\bigskip
Technique!!...
%and finally
\eject
!endverbatim
This markup definitely reflects structure,
I also use \cs{beginquote}, \cs{endquote}, in order to quote
from literature within the text.
\subsubhead{Exercise and answer} markup macros have been used
for formatting the \TeX book, and are provided in manmac.tex.
For the answer the writing to a file verbatim has been used,
in order to typeset the answer at a place different
from the occurrence in the compuscript.
When proofing the answer can be set right after the exercise.
Active documents, aha! The markup reads as follows.
\example{\TB~p.~4}
\beginverbatim
\exercise What do you think happens when
   you type four hyphens in a row?
\answer You get em-dash and hyphen
(----), which looks awful.
\par%or blank line
!endverbatim
\subsubhead{Math} from plain can be enriched by
automatic numbering and cross-referencing,
as well as with special symbols.
The cross-referencing has been worked out in `BLUe's
Cross-Referencing.'\ftn{For special symbols use AMS' fonts.}
\example{From BLUe's Cross-Referencing}
Look at the places where normally the explicit equation number
should be supplied. \cs{ref} optionally followed by |\<name>|
is all what has to be inserted. For referencing provide
\cs{crsref} |\<name>|. That is all, a compatible extension.
I love it.\ftn{It is true
   that forward references need the insertion of the number
   in the script manually before the ultimate processing.}
\beginverbatim
\item{--} Labeled 1-line
$$\sin2x=2\sin x\, \cos x  \eqno\ref\cgla$$
\item{--} Three lines, second flushed right
(relevant for 2-column printing)
$$\displaylines{F(z)=
a_0+{a_1\over z}+{a_2\over z^2}+\cdots
   +{a_{n-1}\over z^{n-1}}+R_n(z),\cr
           \hfill n=1,2,\dots\,,\cr
\hfill F(z)\sim\sum_{n=0}^\infty a_nz^{-n},
      \quad z\to\infty\hfill\ref\cglb\cr}$$
\item{--} Two lines aligned, with middle
          labeling
$$\eqalign{\cos2x&=2\cos^2x-1\cr
                 &=\cos^2x-\sin^2x\cr}
                                \eqno\ref$$
\item{--} Two lignes aligned, with labeling
          per line
$$\eqalignno{
\cosh2x&=2\cosh^2x-1&          \ref\cgld\cr
       &=\cosh^2x+\sinh^2x\cr}$$
and citations \cite\cgla, \cite\cglb,
\cite\cgld
!endverbatim
\noindent with result
\medskip
\item{--} Labeled 1-line
$$\sin2x=2\sin x\, \cos x  \eqno\ref\cgla$$
\item{--} Three lines, second flushed right
(relevant for 2-column printing)
$$\displaylines{F(z)=
a_0+{a_1\over z}+{a_2\over z^2}+\cdots
   +{a_{n-1}\over z^{n-1}}+R_n(z),\cr
           \hfill n=1,2,\dots\,,\cr
\hfill F(z)\sim\sum_{n=0}^\infty a_nz^{-n},
      \quad z\to\infty\hfill\ref\cglb\cr}$$
\item{--} Two lines aligned, with middle
          labeling
$$\eqalign{\cos2x&=2\cos^2x-1\cr
                 &=\cos^2x-\sin^2x\cr}
                                \eqno\ref$$
\item{--} Two lignes aligned, with labeling
          per line
$$\eqalignno{
\cosh2x&=2\cosh^2x-1&          \ref\cgld\cr
       &=\cosh^2x+\sinh^2x\cr}$$
\noindent and citations \cite\cgla, \cite\cglb,
\cite\cgld.
When I worked on `Math into BLUes,' I  considered it handy
to extend \cs{eqalign} with a repetitive template such that
more than one alignment point is accounted for.
This extension has been incorporated in blue.fmt.
At that time I was quite surprised that
\LaTeX{} does not allow as such
the inclusion of the \cs{eqalign} family of macros,
because \cs{centering} is used in \LaTeX{} for a different purpose,
that is, it is not a glue quantity.
\subsubhead{Tables} can be marked up by
\cs{halign},\ftn{Or its transpose \cs{valign}.}
 or the alignment display.  %The latter is used at the end
%of the article. Part of its markup is shown below.
The latter is heavily used in manmac at a lower level.
\example{Table of markup tags}
\begindisplay\displayindent\parindent
Tags  &Reg script\quad&Tansparencies\cr
\noalign{\vskip.5ex\hrule\vskip1ex}
\multispan3{Token variables\hfil}\cr
\cs{title}         &+      &+\cr
%et cetera
\enddisplay
is obtained via
\beginverbatim
\begindisplay\displayindent\parindent
Tags  &Ref script\quad&Tansparencies\cr
\noalign{\vskip.5ex\hrule\vskip1ex}
\multispan3{Token variables\hfil}\cr
\cs{title}         &+      &+\cr
%et cetera
\enddisplay
!endverbatim
\noindent Note the special use of `option's after \cs{begindisplay}
till the end of line. The invoke of \cs{halign} and its template
are hidden by \cs{begindisplay}.
Hardly no overhead, neat! Under the hood use has
been made of the powerful, but much neglected, alignment display.
 Font charts are special. I find Knuth's
\cs{beginchart} \cs{endchart} macros unsurpassed.
\example{Font charts, \TB{} Appendix~F, 427, 428, 429, 430, 431, 432}
\beginverbatim
   \beginchart\tenrm
      %\tenit, \tentt, or ... (see above)
      \normalchart
   \endchart
!endverbatim
\noindent Is that all?!? What does the result look like?
Yes, definitely,
manmac reflects descriptive markup, to say the least.
For the result in print open the \TeX book.
At a higher level I find the bordered table approach handy,
because this allows me to treat the border separately,
next to abstract from the row and column separators.
At the outer level the logical number of columns and rows
can be easily recognized from the markup.
\example{From `What is \TeX{} etc.'}
\beginverbatim
\def\data{11\cs12\rs21\cs22}
\item{--} just framed data
   $$\vcenter{\framed
     \beginbtable\data\endbtable}$$
\item{--} add header and rowstubs
  \def\header{\logms2\hfill
                   Header\hfill}
  \def\rowstblst{{$1^{st}$ row}%
               {{$2^{nd}$ row}}}
  $$\vcenter{\beginbtable\data\endbtable}$$
\item{--} add caption and footer,
      vary via dotted lines
  \def\btablecaption{Caption}
  \def\footer{Footer}
  $$\vcenter{\dotruled
    \beginbtable\data\endbtable}$$
\item{--} vary with ruled and framed
  $$\vcenter{\ruled\framed
    \beginbtable\data\endbtable}$$
!endverbatim
\noindent with result
\medskip
{\def\data{11\cs12\rs21\cs22}
\item{--} just framed data
   $$\vcenter{\framed
     \beginbtable\data\endbtable}$$
\item{--} add header and rowstubs
  \def\header{\logms2\hfill
                   Header\hfill}
  \def\rowstblst{{$1^{st}$ row}%
               {{$2^{nd}$ row}}}
  $$\vcenter{\beginbtable\data\endbtable}$$
\item{--} add caption and footer,
      vary via dotted lines
  \def\btablecaption{Caption}
  \def\footer{Footer}
  $$\vcenter{\dotruled
    \beginbtable\data\endbtable}$$
\item{--} vary with ruled and framed
  $$\vcenter{\ruled\framed
    \beginbtable\data\endbtable}$$
\noindent{\bf Remark.} Tables which are larger
than a column (in \cs{twocol}) can be put in an Appendix
for the moment, with \cs{onecol} switched on.
No \cs{multicol} output routine of yet, which allows
$1\leftrightarrow2$  column switching
and vice versa anywhere on the page.
Tables which extend the page with header automatically repeated
on every top of the page are not treated. Consult for those
the CTAN.\ftn{I would implement this by modifying the OTR
   such that the toks variable tableheader will be inserted.
   If empty nothing will appear.}
Verbatim in tables is not special either. The only thing that
matters is that the table contents is processed on the fly.
That comes down usually to processing via two-part macros, definitely
not storing the contents.
\subsubhead{Pictures} have to do with
 place within context\ftn{A general aspect for
     document elements which don't look nice over  a page break,
     or have to be formatted with something special around
     them.}
and with the picture proper.
 For graphs one can use plain's turtle graphics,
 manmac's line-diagrams, or
 gkpmac.\ftn{The macros for graphics are not robust.
   Vectors of length 0 go wrong (no arrowheads as such).
   Placement of arrowheads at the end of \cs{squines}
   needs trial-and-error adjustments. Inconsistent
   with \LaTeX{} is the reference point of
   \cs{makebox}.
   In `BLUe's Graphs' I have discussed these issues.
   I also added \cs{framebox}.}
\subsubhead{Plain's} boxes approach.
\example{Scheme via plain, from AMS BLUes}
\beginverbatim
\def\element#1{\hbox to15ex{\hss#1\hss}}
$$\hbox{\vbox{\lineskip.5\lineskip
\element{author(\TeX)}
\element{$\downarrow$}
\element{\tt amsppt.sty}
\element{$\downarrow$}
\element{\tt amstex.tex}
\element{$\downarrow$}
\element{\TeX}
}\qquad\qquad\qquad\qquad
\vbox{\lineskip.5\lineskip
\element{ author(\LaTeX)}
\element{$\downarrow$}
\element{\tt amsart.sty}
\element{$\downarrow$}
\element{\llap{{\tt amstex.sty}$\,%
\rightarrow\;$}\LaTeX}
\element{$\downarrow$}
\element{\TeX}
!endverbatim
with result
\def\element#1{\hbox to15ex{\hss#1\hss}}
\hbox{\vbox{\lineskip.5\lineskip
\element{ author(\TeX)}
\element{$\downarrow$}
\element{\tt amsppt.sty}
\element{$\downarrow$}
\element{\tt amstex.tex}
\element{$\downarrow$}
\element{\TeX}
}\qquad\qquad\qquad\qquad\vbox{\lineskip.5\lineskip
\element{ author(\LaTeX)}
\element{$\downarrow$}
\element{\tt amsart.sty}
\element{$\downarrow$}
\element{\llap{{\tt amstex.sty}$\,%
\rightarrow\;$}\LaTeX}
\element{$\downarrow$}
\element{\TeX}
\subsubhead{Manmac's} \cs{makeblankbox} approach.
\example{From Manmac BLUes}
\beginverbatim
$$\hbox{\vbox{%
\element{\fbox{amsppt.sty}}
\vconnector
\element{\fbox{amstex.tex}}
\vconnector
\element{\fbox{\TeX}}
}\qquad\qquad\qquad\vbox{%
\element{\fbox{amsart.sty}}
\vconnector
\element{\llap{\fbox{amstex.sty}---}
 \fbox{\LaTeX}}
\vconnector
\element{\fbox{\TeX}}
!endverbatim
\noindent I obtained
{\def\strut{\vrule height2.5ex depth1ex width0pt}
\def\fbox#1{\setbox0\hbox{\strut
 $\;$#1$\,$}\leavevmode\rlap{\copy0}%
 \makelightbox}
\def\element#1{\hbox to15ex{\hss#1\hss}}
\def\vconnector{\element{\strut\vrule}}
$$\hbox{\vbox{%
\element{\fbox{amsppt.sty}}
\vconnector
\element{\fbox{amstex.tex}}
\vconnector
\element{\fbox{\TeX}}}
\qquad\qquad\qquad
\vbox{%
\element{\fbox{amsart.sty}}
\vconnector
\element{\llap{\fbox{amstex.sty}---}%
 \fbox{\LaTeX}}
\vconnector
\element{\fbox{\TeX}}}
with the auxiliaries\ftn{\cs{makeblankbox} is used in \TB{}
   in exercise 11.5, for the skyline of words.}
\beginverbatim
\def\strut{\vrule height2.5ex depth1ex
 width0pt}
\def\fbox#1{\setbox0\hbox{\strut
 $\;$#1$\,$}\leavevmode\rlap{\copy0}%
 \makelightbox}%See appendix
\def\element#1{\hbox to15ex{\hss#1\hss}}
\def\vconnector{\element{\strut\vrule}} .
!endverbatim
{\bf gkpmac} the generic solution.
gkpmac provides \LaTeX's picture environment in essence.
My added value to this collection is that I provided the possibility
for options via my \cs{this}\dots\ and \cs{every}\dots.
gkp's \cs{beginpicture} remains accessible
because I renamed it with prefix gkp into \cs{gkpbeginpicture}.
\example{Lines, text and circle/disk}
\beginverbatim
\thispicture{\unitlength=1cm
   \xdim{7}\ydim{3}
   \yoffset{-1.5}
\beginpicture
\line(1,1){1}
\line(1,-1){1}
\makebox(2,0){Beer}
\disk{.25}\quad
\circle{.5}
\endpicture
!endverbatim
with result\ftn{Note the omission of \cs{put}-s in this simple case.}
$$\unitlength=1cm
\gkpbeginpicture(7, 3)(0,-1.5)
\line(1,1){1}
\line(1,-1){1}
\makebox(2,0){Beer}
\disk{.25}\quad
        \circle{.5}
\gkpendpicture$$
\noindent
The new \cs{beginpicture} with uncoupled coordinates and
offsets allows an interesting possibility to store a picture
and when using it  to provide it with the actual
values of \cs{unitlength} and the offsets.
An invocation comes down to
\beginverbatim
\<name>pic
!endverbatim
eventually preceded by \cs{this\-pic\-ture} or
\cs{every\-pic\-ture}.
\beginverbatim
\thispicture{\unitlength=<dimen value>
   \xoffset{<number>}\yoffset{<number>}}
!endverbatim
\noindent The database should contain the picture
\beginverbatim
\def\<name>pic
{\bgroup%defaults
   \unitlength=<dim value>
   \xdim{<number>}\ydim{<number>}
\beginpicture
   ....%picture descriptions
\endpicture
\egroup}
!endverbatim
\noindent with
\beginverbatim
\def\beginpicture{\bgroup
   \the\everypicture\the\thispicture
   \gkpbeginpicture(\the\xdim,\the\ydim)%
         (\the\xoffset,\the\yoffset)}
\def\endpicture{\gkpendpicture\egroup
   \global\thispicture{}}
!endverbatim
\noindent A gkpmac compatible extension.\ftn{The name changes
   with prefix gkp
   have been incorporated in the gkppic file.}
The size and the offset have been decoupled
from the picture. Interesting, very interesting.
The basis for a database of pictures has been laid.
The \TeX nique of selective loading similar to the one
used with bibliography entries can be applied too.
The same structural approach has been followed.
\subsubhead{Selecting and loading.}
Imagine you have built up  a number of pictures
each consisting of a reasonable amount of picture instructions.
Then you can either
\bitem input the whole file and use only
       those entries needed, via \cs{loadallpictures}, and not via
      \cs{input pic.dat} because
       the common aspects have to be activated, or
\bitem load selectively only those needed.
\smallbreak
I have used  selective loading
before in `BLUe's Bibliography,' and it can be used easily for
this selective loading of pictures from a picturebase.
The basic observation is that
specified control sequences are redefined. This in contrast
with the application of a format or style file where the file
is not loaded when some names are already available, to
prevent double loading.
My use is the opposite. I specify the names I like to be selected.
\example{Oval via gkpmac}
\beginverbatim
$$\originpic$$
%with
\def\originpic
{\bgroup\unitlength1cm
   \xdim{3}\ydim{1}
\beginpicture
  \put(1,.5){\oval(2,1)}
  \put(0,0){\markorigin}
\endpicture
\egroup}
!endverbatim
 with result
\def\originpic
{\bgroup\unitlength1cm
 \xdim{3}\ydim{1}
\beginpicture
  \put(1,.5){\oval(2,1)}
  \put(0,0){\markorigin}
\endpicture
\egroup}
$$\originpic$$
Note that the placement within context is handled by the math display
which seems enough for me to refrain from the use of \cs{prepicture}
and \cs{postpicture}.
The following examples assume that the pictures are available in the
picture database. No need for \cs{thispicture} but it can be used
to provide user guidance.
\example{Vectorbundle via gkpmac}
\beginverbatim
\vectorbundle
!endverbatim
with result
\vectorbundlepic
\example{Flow chart (Furuta)}
\beginverbatim
\furutapic
!endverbatim
with result
\noindent
\furutapic
Remark. From \LaTeX's line fonts it can be seen that the arrowheads
are designed to match seemless with line elements to form vectors.
When direction indicators are needed along with curves I use
PMA---poor man's arrowheads, built from a small \cs{hrule}
and \cs{vrule}---as can be seen from the accompanying flow chart.
In general I think that arrowheads with the heart of the arrow as reference
point would be  useful.
\example{Pie-chart}
A more cumbersome use of \cs{squine} has been realized in the following
`circle.' This figure was easily transcribed from my earlier \LaTeX{} variant
based on the use of \cs{bezier}.
\beginverbatim
\ntgpiepic
!endverbatim
with result
\ntgpiepic
\subsubhead{Verbatim} macros are needed by authors who write in
\AllTeX{} about \AllTeX.\ftn{Erik Frambach communicated that they are
   also used for programs written in other languages than \AllTeX.}
Usually we distinguish in-line versus dislay verbatim facilities.
Furthermore we need file verbatim inclusion, and semi-transparent
verbatim, meaning that an escape character is available
allowing us to do whatever we wish.\ftn{Not to forget to end the
    verbatim.}
In blue.fmt I included the macros which emerged from `BLUe's Verbatim.'
\example{From BLUe's Verbatim}
Important functionalities are shown below
\thisverbatim{\catcode`\|=12 \let\-=-
              \catcode`\!=12
              \catcode`\<=12
              \catcode`\-=0 }
\beginverbatim
%1. To handle via default ! escape char
%   -- other fonts
%   -- footnotes
%   -- emc (enable metacode)
\thisverbatim={\emc
   \def\ftxt{Footnote text typed in
             on more than one line.}}
\beginverbatim
Some <meta code> and
blah, blah, ...    !it
Now text in italics!tt
and back again in tt
footnote!footnote!bgroup!rm!ftxt!egroup
!endverbatim
%2. To handle numbering and
%   verbatim file inclusion
\everyverbatim={\numvrb}
\thisverbatim={%
   \catcode`\!=12 % default escape char
   \catcode`\|=12 % used for inline verb
   \input vrb.tex
   \catcode`\*=0 }% to terminate
\beginverbatim
Extras after file
*endverbatim
%3. To restart (line)numbers
\thisverbatim={\vrblin0 }%Or \numvrb
\beginverbatim
Just some text with
ligatures such as ?` switched off (TB 381).
line numbers restarted via
\vrblin0 (in general via \numvrb).
After two blank lines.
!endverbatim
%4. Inline verbatim via minimal tags
%Minimal | tag for inline verbatim
\makeactive\|
\def|{\bgroup\setupverbatim
        \the\everyverbatim
        \the\thisverbatim
        \def|{\egroup\thisverbatim{}}}
\thisverbatim{\emc}
Before |<inline verbatim>| after.
-endverbatim
\noindent Remarks. The \vrt-definition for the minimal tag has also been
included in blue.fmt. Beware, don't use \vrt\ as such for
the symbol, but mark up with \cs{vrt} when just the symbol is needed.
The escape character in verbatim is very convenient. I used  this
for subscripting in verbatims, and for selective numbering
of the macros in  verbatim listings, meaning that code parts can
start with a nice number, like 100, 200 and so on.
This can be done by including
\beginquote
\%|<escape char>vrblin|= 100, respectively\\
\%|<escape char>vrblin|= 200.
\endquote
I refrained for the moment to provide Knuth's way of handling
of tab characters, that is a variable number of spaces.
See for that \TB, p.391.
\subsubhead{Special texts} are for example programs.
Manmac's syntax macros can be used to format the syntax rules.
I love those macros, they guide you towards
natural markup. Superb again!
\example{\TB~p.~268}
\beginverbatim
\beginsyntax
<unit of measure>\is<optional spaces>%
                    <internal unit>
\alt<optional {\tt true}><physical unit>
\endsyntax
!endverbatim
The markup of the top level part of the syntax chart
for a blue script reads
\thisverbatim{\catcode`\|=12
              \unmc}
\beginverbatim
\beginsyntax
<blue script>\is|\input|\thinspace%
        blue.fmt\thinspace<scripts>
<scripts>\is<script>\alt<script><scripts>
<script>\is<preliminaries>
 \quad<script options>
 \quad|\beginscript|
  \qquad<sections>
 \quad|\endscript|
\endsyntax
!endverbatim
\subhead{Handy} is manmac's possibility for
the shipping out of selected pages.
All what has to be done is to supply the page numbers
in the file pages.tex, each number on a separate line,
and in increasing order.
Also handy are tugboat.cmn's abbreviations and  macros for logos,
to enhance consistency.
\subhead{Syntax diagrams.} The template given earlier can more
formally be described via the following syntax diagrams,\ftn{I
   refrained from more elaborated notational schems, for example
   those used in the SGML world. The reason is simply that
   they are not needed to convey the idea. The basics of
   Backus-Nauer notation with some comments will do, similar to
   the way the syntax diagrams are used in the \TeX book.}
in the spirit of chapter 24 of the \TeX book.
IM copy denotes script parts
marked up with lower level \TeX{} markup.
\beginsyntax
<blue script>\is|\input|\thinspace%
            blue.fmt\thinspace<scripts>
<scripts>\is<script>\alt<script><scripts>
<script>\is<preliminaries>
 \quad<script options>
 \quad|\beginscript|
  \qquad<sections>
 \quad|\endscript|
\endsyntax
with refinements
\beginsyntax
<preliminaries>\is|\title|
 \alt|\transparencies||\title|
 \alt|\<yours>||\title|
 \alt<preliminaries><preliminary>
<preliminary>\is|\subtitle|\alt|\issue|
 \alt|\input|\alt<TeX code>
 \alt|\author||\address|%
      |\netaddress|\ftn{Defaults.}
 \alt|\keywords|\alt|\abstract|
 \alt|\contents|\alt|\references|
 \alt|\pictures|
<script options>\is<empty>
 \alt|\everyscript|\alt|\thisscript|
 \alt|\everyscript||\thisscript|
\endsyntax
The sections consist of the tree of head
structures with IM copy, inner marked up copy proper.
\beginsyntax
<sections>\is<empty>\alt<section><sections>
<section>\is|\head|<paragraphs>
<paragraphs>\is<empty>
   \alt<paragraph><paragraphs>
<paragraph>\is<ssections>\alt<IM copy>
\endsyntax
The above is repeated for sub(sub)sections.
\beginsyntax
<ssections>\is<empty>\alt<ssection><ssections>
<ssection>\is|\subhead|<sparagraphs>
<sparagraphs>\is<sssections>\alt<IM copy>
<sssections>\is<empty>\alt<sssection><sssections>
<sssection>\is|\subsubhead|<IM copy>
\endsyntax
\subhead{The semantics} is natural.
It is sensible to provide for one set of keywords per script.
Ditto for abstract and contents.
More than one author has to be handled within
\cs{author},\cs{address}, and \cs{netaddress}.\ftn{Of course
   I could have provided for an array of authors, but hopefully
   this will do.}
\subhead{Manmac one-column} specifics are
essentially the mathdemo variants and the font charts.\ftn{Well,
   \cs{setcornerrules} makes use of the width of the page.
   So my conclusion is that manmac is not restricted to
   one-column format!}\\
\rightline{Surprisingly few isn't it?}
\head{Coding}
In this section just a glimpse of the coding is
supplied---the better ones---as appetizer.
For inner markup coding consult the references on the issue.
\subhead{Coding conventions.}
I assumed  as little  \TeX{} knowledge as possible for the outer level
markup.
I value a discipline of coding highly.
An example of this is not to mix up
token variables with definitions.
The choosing of names is important too, if not for
enhancing mnemonics.
In pursuit of Knuth I favour |\begin!-<tag>| end |\end!-<tag>|
pairs.\ftn{In earlier versions I was not so clear about
   this and used |\<tag>|, and |\end<tag>| pairs. Unnecessary
   `Syntactic Sugar,' which is confusing.}
I like to program in paradigms and code these systematically
given the language at hand, supported by templates, to be used
over and over.\ftn{An example is the use of the FIFO paradigm.}
Abstraction is important too. I have avoided the use
of tags like \cs{do\-cu\-ment}, because that is not what
it is about.
I composed from the rootname |<tag>|
names with prefixes `begin,' `this,'
`every', `pasteup', `pre', `post' or `end,' and
names with postfix  `box,' and `name.'
The latter stores a language dependent name.
Token variables are used to convey information.\ftn{In most
   cases an author won't realize that there is {\em no\/}
   difference in the
   markup for the invocation of a macro with one
   argument and the assignment to a token variable.
      In general using the token variable is different from
                 using the replacement text of a definition.
   \cs{the}\cs{\hbox{$\langle tag\rangle$}}
   versus without \cs{the} for the definition.}
Other control sequences have a rootname,
which can be used either as a macro with a parameter,
or as a `two-part'  macro, the latter
with prefixes |begin|, respectively |end|.
An example is \cs{head}.\ftn{The \cs{head} definitions take as `argument'
   the head title processed on the fly. No token variables for these!}
I'm not  complete with names, for example  |\this!-<tag>|s
are not provided for the rootname |head|.\ftn{For a survey
   of which names are implemented see the tables in the Appendix.}
When there is a need for it, yes, it has been added.
And if you like one of your own, go ahead add one.
For language dependent names I introduced definitions with names
 |\<tag>name|, for example for use within the environments
keywords, abstract and references.\ftn{No parsing of options!}
When I built upon available macros I added an extra level
to comply with my markup conventions. When the original name of the
macro had te be preserved for the outer level I changed
the  name with a  prefix pointing to the source.
\subhead{Beginning of the blue script.}
This is included because it shows
the paste-up of the various title elements, and  the coding
of the mechanism to run various scripts  in one job.
\subsubhead{Design goals} are
\bitem preliminary matter can be provided in any order
\bitem start with a rule in print
\bitem set title in bold
\bitem set authorname indented
\bitem store address information to be set
       at the end
\bitem set keywords in smaller type
\bitem set abstract `narrowed' in smaller type
\bitem set (small) table of contents
\bitem title and issue are used in running headline
\bitem running foot contains  draft, pagenumber
       and copyright.
\smallbreak
\subsubhead{Coding} is as follows
\beginverbatim
\def\beginscript{\lastscript
   \the\everyscript\begingroup
   \the\thisscript
   \hrule\kern2ex\noindent
   {\bf\the\title{\rm\the\subtitle}}
   \medskip\the\author
   \medskip\pasteupkeywords
   \medskip\pasteupabstract
   \medbreak\pasteupcontents
   \smallbreak}
\def\endscript{\makesignature
   \xcol=\maxcols
   \vfil\eject\endgroup\tracingstats1
   \stop\thisscript{}}
%with auxiliaries
\def\pasteupkeywords{\box\keywordsbox}
\def\pasteupabstract{\box\abstractbox}
\def\pasteupcontents{\unvbox\contentsbox}
\def\notlastscript{\global
    \let\stop\relax}
\def\lastscript{\global\let\stop\end}
!endverbatim
The formatting of the |\<name>box|
varies with the |<name>| and is done
as follows
\beginverbatim
\def\beginkeywords{\setbox\keywordsbox
 \vbox\bgroup\subhead{\sevenbf
 \the\keywordsname: }\small}
\def\endkeywords{\egroup}
\def\beginabstract{\setbox\abstractbox
 \vbox\bgroup\centerline{\sevenbf
 \the\abstractname}\beginquote\small}
\def\endabstract{\endquote\egroup}
\def\begincontents{\setbox\contentsbox
 \vbox\bgroup\small\noindent
 \the\contentsname\obeylines}
\def\endcontents{\egroup}
!endverbatim
The above implements the opening of the two-column
result of a blue script.
\subhead{Headers.}
I will design \cs{beginhead}  and \cs{endhead},
next to the minimal markup variant |\head{...}|.\ftn{The subhead
   variants are similar but different in detail. Especially
   in undoing  glue when heads follow-up
   each other without intervening text. Beware!}
\subsubhead{Design goals}  are
\bitem discourage to  set the title alone at the end of a page
\bitem typeset (flexible) vertical space  before (big) and after (med) the titl
\bitem gobble spaces at the beginning of the title
\bitem set the title in bold face (and the current size) unindented
\bitem don't indent the first line of the text after.
\subsubhead{What are the problems?}
The language constructs have to be chosen.
In macro expansion the two-part macro
\TeX nique should be the basis. In blue's format I adopted
therefore the pairs |\begin<tag>| and |\end<tag>|.\ftn{Short
   variants, also called minimal markup, is an anachronism
   with nowadays \AllTeX{} intelligent editors.}
In order to set the title loose from the context
we have to determine values for the amount of glue,
and be sure that it disappears at the top of a page.
Another problem is to prevent  the headtitle alone
to be printed at the bottom of the page. The latter is
related to the `widow-orphan' phenomenon.
The coding solution below has different values for the
parameters as supplied in \cs{beginsection}.
From experiments it turned out that those supplied by Knuth
did not give nice results together with `BLUe's Format.'
Too much glue was inserted for my taste.
The idea of a |\pre<tag>| and |\post<tag>| is rather general.
It parameterizes the placement of a document element
within context. Important! Whether you talk about titles,
displays, tables, graphs, or you name it.\ftn{It is important
   to pay good attention to these constructs because
   they are  dominant on each page, they contribute highly to the
   look-and-feel.}
\beginverbatim
\def\beginhead{\the\prehead\bgroup\headfont}
\def\endhead{\egroup\the\posthead}
%with assignments to the token variables
\prehead{\vskip0pt plus5ex
   \penalty-250\vskip0pt plus1ex
   \bigskip\noindent}
\posthead{\medskip\nobreak
   \noindent\ignorespaces}
!endverbatim
The paradigm to be learned from this example is that we should
build upon two-part macros.
\subsubhead{Take a deep breath.}
The minimal markup form,\ftn{Yes, I agree minimal markup is not
   much needed anymore with the emerge of \AllTeX{} intelligent
   editors. The coding paradigm is of general interest, though.}
{\em with the same functionality}, reads
\beginverbatim
\def\head#{\beginhead\bgroup
   \aftergroup\endhead
   \afterassignment\ignorespaces
   \let\dummy=}
!endverbatim
\noindent Explanation.
The \# as last character of the parameter text makes that
the opening brace and what follows
is placed after the replacement text of \cs{head}, \TB, p204.
The replacement text starts a group via  \cs{bgroup} and
reads away the opening brace. The latter \TeX nique has been
borrowed from plain's \cs{footnote}, \TB, p.363,
as used at the end of the replacement text of \cs{fo@t}.
Spaces which precede the title are ignored via
the invoke of \cs{ignorespaces} at the right time.
Quite something isn't it?
\subsubhead{Paradigm.}
From the above variant the following
template can be distilled,
to be used with {\em similar\/} two-part
macros for a minimal markup variant
\beginverbatim
\def\<tag>#{\begin<tag>\bgroup
   \aftergroup\end<tag>
   \afterassignment\ignorespaces
   \let\dummy=}
!endverbatim
\subsubhead{Take a deeper breath.}
When using the above template in practice it occurred to me
that \TeX{} could handle that too.
For \cs{head} the desired  functionality is that
\beginverbatim
\onepartmacro{head}
!endverbatim
will build the macro \cs{head} based upon
the two-part macros \cs{beginhead} and \cs{endhead}.
The naive coding\Dash which does not work, but makes
the purpose  clear\Dash reads
\beginverbatim
\def\onepartmacro#1{\def\#1##{\begin#1\bgroup
   \aftergroup\end#1
   \afterassignment\ignorespaces\let\dummy=}}
!endverbatim
Several desired functionalities have to be coded
in \TeX's peculiar way, with as result
\beginverbatim
\def\onepartmacro#1{{
\ea\let\csname begin#1\endcsname=0
\ea\let\csname end#1\endcsname=0
\toks0={\bgroup\aftergroup}
\toks1={\afterassignment\ignorespaces
        \let\dummy=}
\ea\xdef\csname#1\endcsname##{%
  \csname begin#1\endcsname
  \the\toks0 \csname end#1\endcsname
  \the\toks1 }}}
!endverbatim
Explanation. A lot of the trickery about partial
expansion, \TB, p.216, has been applied.
First a \cs{csname} is used to compose the names from
the rootname.
Second the one-part macro is defined via an \cs{xdef}
with suppression of  the expansion of the
|\begin<name>|, |\end<name>|, and the other tags.
An \cs{xdef} is used to keep the auxiliaries local.
The first is handled via the property that a control sequence
equal to a character is not expanded.
The second is handled by the property that a \cs{the}
on a token list is a one-step expansion.
It is left as an exercise for the reader to
do the transformation process step by step.
\subsubhead{Some remarks.}
Note that I did not gobble spaces at the end of the title argument.
In the headtitle they won't harm because
the title is set on a line of its own.\ftn{In the other heads
   the space at the end of the argument can be there on purpose.
   Watch out!}
Processing a title on the fly is needed for verbatims
and in general when we don't like the `eyes' of \TeX{}
to set catcodes differently from those the script expects
when digested in the `stomach.'
\subsubhead{Disclaimer.}
The above template does not work for general two-part macros,
for example not for
\bitem verbatims\Dash esc char interference\Dash
\bitem btable\Dash group and \cs{halign} interference
\bitem `argument' is used twice, for example
as in \cs{references}.
\smallskip
For those cases I fell back on a macro with parameter.
To keep the flavour of the two-part macro in there, I used
|\end<tag>| as endseparator.
\subhead{AN-items.}
A numbered list can be obtained by changing \cs{item} and \cs{textindent}.
I chose for the compatible extension to add  macros on top in
Knuth's spirit.
No \cs{beginitem} and \cs{enditem} are needed because
I use the paradigm that the first invocation redefines itself
next to doing what it is supposed to do. That all within a scope where
\cs{smallbreak} has been redefined to end the scope and invoke
the original \cs{smallbreak}.
\beginquote
\cs{nitem}, numbered items\\
\cs{aitem}, lettered items.
\endquote
\beginverbatim
\newcount\itemno
%Automatically numbered items
\def\nitem{\bgroup
   \def\nitem{\advance\itemno1
      \item{\number\itemno.}}
   \def\smallbreak{\endgraf\egroup
      \smallbreak}
   \let\smallskip\smallbreak
   \nitem}
%Automatically lettered items
\def\aitem{\bgroup\itemno96 %lc
                         %64 uppercase
   \def\aitem{\advance\itemcnt1
      \item{\char\itemcnt.}}
   \def\smallbreak{\endgraf\egroup
      \smallbreak}
   \let\smallskip\smallbreak
   \aitem}
%an extra markup tag for ending
\def\endlist{\endgraf\egroup}
   %No \smallskip on purpose
!endverbatim
Numbered or lettered
variants for \cs{itemitem} are left
as an exercise for the reader.
\subhead{Selective loading of pictures.}
This is simpler at the moment than
the handling of references.
The specification comes down to
\bitem load  specified definitions
\bitem prevent side-effects (work within a group)
\bitem separate typesetting from paste-up
\bitem cope in an elegant way with
       typos of the names.
\smallbreak
\noindent At the heart of the coding lies the observation  that
specified control sequences are redefined. This in contrast
with the common use to prevent a format or style file to be overwritten.
My use is the opposite. I specify the names I like to be selected.
\thisverbatim{\catcode`~=14
   \catcode`!=12
   \catcode`+=0 }
\beginverbatim
\def\beginpictures#1\endpictures{%
  \bgroup\let\commonforpic=x
  \def\process##1{\gdef##1{Picture
  {\tt\string##1} not in database,
  Sorry!\loaderror{Picture}}}%
  \fifo#1\ofif
  \loadselectivefrom{pic.dat}
  \endpictures\commonforpic}
\def\endpictures{\egroup}
\def\pictures#1{\beginpictures#1%
   \endpictures}
\def\loadallpictures{\input pic.dat
   \commonforpic}
\def\loaderror#1{\write16{#1 not in
      database}}
+endverbatim
\noindent Explanation. First, the common name is made known.
Second, via the FIFO paradigm all
supplied names are made known (all equal to the
error message, a  useful fake).
Third, the known names are compared with names occurring in pic.dat
and when in there
the fake is  globally replaced
by the proper picture replacement text.
All is done within a group to prevent side-effects.
At the end the common things have to be invoked.
I also added a macro to load all pictures.
This is a bit special because of the imvocation of the common part.
It is not enough to say \cs{input pic.dat}.
The \cs{loadallpictures} can't be used within \cs{pictures},
just a stand alone command.
When using the above  the most common error was a typo in the name.
This error is spotted and a message is given at the place of the
invocation and in the log file.
The process of formatting continues and that seems nice to me.
Remark. The short variant \cs{pictures} has been coded straight away,
similar to \cs{references}.
In the latter the argument\Dash the list of names\Dash had to be stored
afterall, and with pictures similar things might happen in future.
\subhead{One-pass handling of references.}
The goal is to
\bitem load only specified references from a database
\bitem `store' these
\bitem associate the names with
       their sequence number for cross-referening
\bitem separate typesetting from paste-up
\bitem cope with typos in the names in an elegant way.
\smallbreak \noindent
Various mental steps have to be integrated.
\bitem maintain a database of references
\bitem provide a references macro to perform the
       tasks of selecting and formatting
\bitem specify the list of names
       as argument to the macro\ftn{It is used twice:
           for selecting and formatting.}
\bitem load selectively from the database
\bitem format the selected entries
\bitem prevent side-effects
\bitem redefine the names of the selected
    entries by their number and store these globally
\bitem provide a paste-up tag.
\smallbreak
In the code I have made use of the fifo
paradigm.\ftn{This comes down to that the tokens after \cs{fifo}
   are `process'ed first-in-first-out until the sentinel
   \cs{ofif} is encountered and the processing stops.}
\thisverbatim{\catcode`!=12
   \catcode`~=0 }
\beginverbatim
\def\beginreferences#1\endreferences{%
  \bgroup\def\process##1{\gdef##1{%
Reference {\tt\string##1}
not in database (Sorry!)
  \loaderror{Reference}}}\fifo#1\ofif
  \loadselectivefrom{lit.dat}
  %formatting
  \ifstore\global\setbox\referencesbox=
    \vbox\bgroup\fi
  \bcnt0 \lsams%Default ls
  \the\thisreferences
  \def\process##1{\ls{##1}
    \xdef##1{\the\bcnt}}
  \fifo#1\ofif
  \endgraf\endreferences
\def\endreferences{\egroup
  \ifstore\egroup\fi
  \thisreferences{}}
\def\pasteupreferences{%
  \the\prereferences
  \unvbox\referencesbox
  \the\postreferences}
\prereferences{\head{\the
  \referencesname}}
\postreferences{}
\def\references#1{\beginreferences#1
   \endreferences}
~endverbatim
Remark. The short variant \cs{references} has been coded straight away,
because the argument\Dash the list of names\Dash had to be stored
afterall.
The selective loading is hinted at in \TB, p382,
Selective loading of macros.
The idea is to consider \cs{def} as the name
of a macro with as argument the name of
the entry to be considered.
Remind that lit.dat contains entries of the form
\beginverbatim
\def\<name>{...}
!endverbatim
Then if the name is not yet known the replacement
text is gobbled otherwise it is stored
under its name. Neat isn't it.
\beginverbatim
%Load specified entries only
\def\gobble#1{}
\def\loadselectivefrom#1{\let\x\def
 %or any name for x
 \x\def##1{\ifx##1\undefined\ea\gobble
         \else\ea\gdef\ea##1\fi}
 \input #1
 \let\def\x}
!endverbatim
Remark. Generally selective loading is applied
to prevent double  loading. A check is made whether
the control sequence already exists and if so nothing is done.
I needed the opposite.
\head{Practicalities}
When a script is in the proofing phase we have to get rid of overfull
and underfull hboxes and vboxes. A pragmatic approach has been
summarized lucidly by Phil Taylor, 1993. To remind you overfull hboxes
come usually from words which extend the line boundaries.
\subhead{How to get rid of the overfull hboxes?}
In writing in
\AllTeX{} about \AllTeX---especially in narrow columns---long control
sequence names give problems. Generally\ftn{See for manmac
\TB, p.414} the hyphenation algorithm
is not in action for those `words.' Let us assume
that these words are set in \cs{tentt}, then we can activate hyphenation
and allow for some room via
\beginverbatim
\hyphenchar\tentt='055
%and room for example via
\tolerance500
\hbadness=499
\hfuzz=5pt
!endverbatim
\subhead{How to get rid of underfull hboxes and vboxes?}
These generally require adjustments of the script.
Underfull vboxes usually come from boxes which are not split over
page boundaries. Then we can decide to let these elements float
\beginverbatim
\topinsert%or \midinsert
<box material>
\endinsert
!endverbatim
\noindent or otherwise adjust  or rewrite the script.
\subhead{Use of memory.}
When I run plain alone I get as report that some
5K memory locations are used. Together with blue.fmt it has
increased to a 20K. Loading of all pictures requires another
10K, and when the memory intensive \cs{ntgpiepic} is run
add 10K to that.
This makes some 40K alone for a document with just the pie-chart.
Loading the complete lit.dat requires 40K. (Some 250 annotated
references.)
Running `What is \TeX{} and \MF{} all about?' which has 125 references
and  memory intensive pictures,
yielded `TeX capacity exceeded.'
That is more than 65.5K was used.
Run with the \cs{storefalse}
option for the references yielded a memory use of 65K (out of 65.5K).
Nobody will be surprised by the conclusion from the above
memory use
\beginquote
when the path of active documents is followed
definitely a big\TeX{} is needed.
\endquote
\head{What not?}
blue.fmt does not deal with hyphenations and font selection.
IMHO font selection should be hidden. A user should generally not
bother about fonts.
Moreover, we should not suffer from `fontitus,'
meaning use as few  fonts as possible. No virtual fonts have
been touched upon either.
It also lacks a fully automated indexing tool.
Furthermore, the use of encapsulated \PS{} has not been treated,
nor macros to handle tables larger than the page size.
\head{Software engineering}
What about correctness, robustness and the like?
\subhead{Correctness.}
Because of the bias of manmac,
correctness is promising to start with.\ftn{The bad news is that
   because I worked on the fly earlier tests have not
   been repeated after every change or modification.}
Similar for gkppic, because it emerged from gkpmac.
blue.fmt has been used for my BLUes and BLUe's series in \cs{twocol},
and for the accompanying transparencies. The latter uses \cs{onecol}.
My `Table Diversions' is  in \cs{onecol} and came out fine.
This article did also run in \cs{onecol}, and as expected more pages
emerged, because of the verbatims which were typed in
with the columnsize of two columns on a page in mind. Moreover these
items did not center, again as expected.\ftn{The pictures can be
   centered independently from one-\-column or two-\-column formatting
   by enclosing them within double \$\$,  a math display.}
\subhead{Robustness.} About this issue I can be very short.
Hardly none of this!
However, I  implemented some sort of  test for the list of
names to occur in the literature database.
It is very frustrating especially with such high-level commands
like |\references{...}| when it does not work and \TeX nical messages
leave you staring at the screen puzzled.
Now I write the message `Name not in database (Sorry!)'
and continue with the good names.
Similar for pictures.
I don't have  an elegant error-handling  when you
forget the backslash of the name. So don't forget the backslash.\ftn{The
   \TeX{} error message shows fifo with the first letter
   of the name separated from the rest. Now you know that the letters
   are processed and not the name.
   Knuth has shown in the \TeX book how to prevent a complex system
   to paralyse you: run the system with the errors you are
   likely to commit and watch how it reacts.}
Because of \TeX's flexibility and redefinition possibilities it is
far too complex for me, to account for robustness in general,
whatever that means
precisely.\ftn{In my numerical software years robustness
   was used to denote the ability of software to react gracefully
   in situations it was not designed for.}
\subhead{Flexibility.}
There are several levels of flexibility.
The first one is to use blue.fmt as-is and complement it with some
macros of your own in the same spirit, hopefully.
One can also customize the macros, or  take whatever
is needed from this collection and add these to your favourite format.
I envision this to happen for the inner level markup macros.
The functionality of `options' has been provided for via
token variables |\this<tag>| and |\every<tag>|.
Also the placement within context of script elements
has been parameterized via |\pre<name>| and |\post<name>|.
The few language dependent names have their name too.
The access to lower levels is there, which means that you can build
from the parts the mods you like.\ftn{The danger of flexibility is
   that confusing variants of blue.fmt will emerge.
   So be it. THE blue.fmt is the one supported by me.}
When this article matured the table of contents grew too big.
I could simply move \cs{contents} and its argument
to an Appendix. \cs{beginscript} handled the unfilled
\cs{contentsbox}, and who cares about that?
\subhead{Consistency.} Definitely, I'm not complete, but strove after
consistency for the outer level markup.
Consistency for the inner level macros suffer from the inherent weakness
of the various sources with their own naming conventions and the like.
\subhead{Orthogonality.} By this we mean that for example with every
two-part macro we can associate a one-part macro, and that we can
combine things arbitrarily.
With respect to the first I did not provide it in full generality.
The one-part minimal variant does not always process the
`argument' on the fly. See the tables in the appendix for a survey.
\subhead{Documentation.} This is a bit peculiar.
Main issues are documented in specific articles on the issue.
A book `Publishing with \TeX' is in progress.
\head{The world outside}
Sooner or later the script will be published
generally first as an article,
and when successful,  collected works will
end up as a book.
In both cases the author has to adapt the marked up compuscript
in agreement with the formatter in use by the publisher,
if not to adapt the contents  because of the increased insight :-)).
I considered the following ways to adapt the compuscript.
\beginquote
When inner markup macros are allowed, even\-tu\-ally
with adap\-ted replacement texts
because of the dif\-fe\-rent look-and-feel in print,
then only the outer markup commands have to be changed.
\endquote
Of course the inner level macros have to be separated from
blue.fmt and delivered to the publisher along with the copy.\ftn{It
   is always good to first simplify the format of the publisher
   into a preprint form. Hopefully, the publisher has provided
   this form already.
   This alleviates the task of adaptation.}
When the publishing of a book is what you are after,
then contact a publisher as soon as possible.
This slogan is often heard and has so its difficulties
for a beginning author.
For the publisher it is difficult to judge
whether a real project will emerge,
whether the book will ever see the light of day.
For the author it is difficult to state what he is up to,
because he does not yet have some sort of proofs.
Because of these uncertainties the author can best rely on his
context independent format.
The bonus is that he can go with his proofs to any publisher he
considers beneficial.
\subhead{MAPS template.}
As example I treat the adaptation
of an article marked up by blue.fmt tags
into an article marked up in the MAPS.sty, which is \LaTeX{} biased,
to give you an impression of the amount of work.
See the accompanying (simplified) MAPS template.\ftn{I don't have
   times fonts,
   don't use Babel (no dutch.sty,
   nor english.sty),  no fancyheadings,
   don't need
   Mittelbach's multicolumn, nor Sch\"opf's verbatim facilities,
   and the like.}
%\topinsert
\beginverbatim
%Template for MAPS submissions
\documentstyle[twocolumn]{maps}
\begin{document}
\title{...}
\subtitle{...}
\author{...}
\date{}
\maketitle
\begin{abstract}...\end{abstract}
\begin{keywords}...\end{keywords}
%Contents table
%Copy proper
\head{...}%et cetera
%back matter, LaTeX's
\begin{thebibliography}{abcde}
\end{thebibliography}
\end{document}
!endverbatim
%\endinsert
\subhead{blue $\rightarrow$MAPS.}
The adaptation from a blue script towards a MAPS script
comes down to elements as given in the accompanying table.
Special attention deserve globals, because they might
interfere with other submissions. The MAPS style is not of
an active type, and therefore retrograde adaptations are needed.
\midinsert
$$\def\x{\lower1ex\null}
\vruled\fll\def\colsep{$\longrightarrow$}
\def\header{BLUe fmt\cs MAPS sty}
\beginbtable
$\vcenter{\hbox{\tstrut|\begin|}
          \hbox{|\end|\x}}$\ abstract
    \cs$\vcenter{\hbox{|\begin{abstract}|\tstrut}
                 \hbox{|\end{abstract}|\x}}$\rs
$\vcenter{\hbox{|\begin|\tstrut}
          \hbox{|\end|\x}}$\ keywords
    \cs$\vcenter{\hbox{|\begin{keywords}|\tstrut}
                 \hbox{|\end{keywords}|\x}}$\rs
|\beginscript|     \cs|\begin{bijlage}| \rs
|\head|       \cs|\let\head\section|\rs
$\vcenter{\hbox{|\bitem|}
          \hbox{\strut\dots}
          \hbox{|\bitem|\x}}$
    \cs$\vcenter{\hbox{|\begin{itemize}|\tstrut}
                              \hbox{|\item...|\strut}
                       \hbox{|\item...|\strut}
                       \hbox{|\end{itemize}|\x}}$\rs
$\vcenter{\hbox{\tstrut|\begin|}
          \hbox{|\end|\x}}$\ quote
    \cs$\vcenter{\hbox{|\begin{quote}|\tstrut}
                 \hbox{|\end{quote}|\x}}$\rs
|\ftn|       \cs|\footnote|\rs
$\vcenter{\hbox{|\begin|\tstrut}
         \hbox{|<escape char>end|\x}}$\ verbatim
       \cs$\vcenter{\hbox{|\begin{verbatim}|\tstrut}
                    \hbox{|\end{verbatim}|\x}}$\rs
|\endscript|  \cs$\vcenter{\hbox{\tstrut
                       |\end{bijlage}|}
              \hbox{|\end{document}|\x}}$
\endbtable$$
\endinsert
The title block is filled in or simply reused
from the template, no problem at all.
A large table of contents is moved to the back part as an
appendix.
As can be seen from the accompanying table I fell back upon
\LaTeX's bibliography handling.
Special attention requires my use of so-called
`running-in' heads.\ftn{The (sub)subhead title is part of the
    opening sentence.}
Important is that blue's verbatim suite, tables and pictures
can be reused as such, similar to math constructs.
For other things like my way of citation\Dash |<author>|, |<year>|\Dash
no adaptation is needed.
\head{Transparencies}
When a presentation results from your work you need slides
or transparencies.
I use the macros as provided in `BLUe's Transparencies,'
which are also included in blue.fmt.\ftn{Adapted with
    for example more
    consistent markup for headings in comparison to
    `BLUe's Transparencies.'}
Just start with
\cs{transparencies} after \cs{input} blue.fmt.
I adopted the same markup tags as for an article as basis.
This makes that the general design problem is reduced
to mainly the typographical design\Dash
       how it should look in print\Dash
and to the coding.
\subhead{Typographical design.}
Only recently I realized that transparencies are completely different
in look-and-feel from pages in a book. Each transparency is more or
less on its own, and has a strong centered flavour.
Therefore my transparencies are centered, but I made an exception
\bitem \cs{bitem}s
\bitem verbatims, and
\bitem syntax rules.
\smallbreak
These are shifted to the right
to reflect the center idea.
My running heads bear contextual information,
a functionality similar to the running heads of a book.
From each transparency the audience
can see at a glance what the context is.
The running foot takes a number, the date
and the copyright information.
In `BLUe's Transparencies' I have included  pictures showing
the layout of the title transparency and the  regular transparencies.
The latter is reproduced here from the picture base via
\beginverbatim
$$\regulartransparencypic$$
!endverbatim
\noindent with result
\midinsert
$$\regulartransparencypic$$
\endinsert
\subhead{Markup.}
As extras I needed token variables for the
head titles, because I like to suppress the headers
on the transparency proper now and then.
Also needed are things like `continue on the next transparency,'
or just ship out the current transparency.
\subsubhead{Template for transparencies.}
It is  a blue script.
%\topinsert
\beginverbatim
\transparencies %Template script
\title{Manmac BLUes}
\subtitle{---how to set a book via \TeX---}
\contents{User's guide
   Coding
   Modifications
   Conclusions}
\beginscript
\head{Why?}
To return to the roots...
\head{User's Guide}
%Relative page numbering
\pagenumber{U\folio}\pageno=1
\nxttrs    %Next transparency
\subhead{Markup paragraphs}
\continue%Continuation on next
\endscript
!endverbatim
%\endinsert
Remarks. I allowed for the possibility to number transparencies by parts.
For example User's guide and Coding can take their own page numbers.
I find that convenient when I modify transparencies.\ftn{In `BLUe's
   Transparencies' I have paid attention to the maintenance aspects.
   How to modify just a few transparencies, or how to add a couple
   to the collection, without the need to rerun  the total collection.}
\subsubhead{Semantics.}
Special for transparencies is that
each logical unit is either restricted to one
transparency or continues on a few.
\cs{head}---as begin of  a new logical unit----automatically
starts a new transparency.
One also needs a tag \cs{continue}
which prints continue in the right lower corner and
starts a new transparency.
Sometimes I want the head or the subhead  title appear only in the heading
and not in the transparency proper.
This can be done by assigning the title to the token variable, as shown
in the template via \cs{subheadtitle}.
Perhaps confusing are \cs{pageno}, plain's countdef, and my
token variable \cs{pagenumber}. I needed a layer in between
and decided to use a token variable. It contains
the page number or the process to obtain it.
If you want to customize the automatic page numbering  of
transparencies,
remember what is already provided by plain---the countdef
\cs{pageno} to set the value---and my in between token variable
\cs{pagenumber} to prescribe the formatting of the page number.
By the first the proper number can be initialized,
and by the second things like U (User Guide) can be added
to \cs{folio}, for example.
\head{Coding}
The coding reflects the desired functionalities but can be improved with
respect to user friendliness and consistency.
The latter especially with two-part macros as basis.
\subhead{Title page.}
For the data of the title page I adopted the toks variables:
\cs{title}, \cs{subtitle}, \cs{author}, \cs{address}.
\subsubhead{Design goal.}
\bitem allow the same markup as for a regular script
\bitem center the author matter on the top half
\bitem use large fonts
\bitem also typeset a table of contents on the lower half.
\smallbreak
The user must supply  the
contents line-by-line. (\cs{obeylines} is on.)
\subsubhead{Implementation.}
The title page is set by \cs{beginscript}.
\beginverbatim
\def\beginscript{\begincenter\bgroup
   \footline={}\headline={}
\message{\the\title}
   \the\everyscript\the\thisscript
   {\titlefont\the\title}
   \smallskip
   {\the\subtitle}\bigskip
   {\authorfont\the\author}
   \smallskip\the\address
   \bigskip$$\box\contentsbox$$}
\def\endscript{\nxttrs\endcenter\end}
%with
\def\begincenter{\bgroup\smallskip
 \leftskip0pt plus1fil
 \rightskip0pt plus1fil
 \parfillskip0pt\obeylines}
\let\endcenter\egroup
!endverbatim
Explanation. The centering is handled globally.
such that each line will be centered.
Because  of \cs{obeylines} I had to deactivate
\cs{parfillskip}, because each line is now a
paragraph. I also set \cs{parindent} 0pt.
\subsubhead{Contents.}
The table of contents is much more prominent
than for an article. It provides the audience
with an overview of the main issues to be treated.
\subsubhead{The design goals}  are
\bitem provide the main points  as argument
   of `contents' in a natural way, each item per line
\bitem sift the table of contents  to the right
\bitem precede each item  by a bullet
\bitem the font size is parameterized in \cs{large}.
\smallbreak
The above has been reflected in the following
code.
\beginverbatim
\def\begincontents{\global\setbox
   \contentsbox\vbox\bgroup
   \leftskip1.5\leftskiptrs
   \large\baselineskip3.5ex
   \everypar{$\bullet$\enspace}
   \obeylines}
\def\endcontents{\egroup}
!endverbatim
The short form is the general
macro \cs{contents}.
\subhead{Regular transparencies.}
Here we have two aspects: what is needed as markup commands,
and what is done automatically with respect to the header and footer.
\bitem the header and footer are set automatically, with information
        distilled from the structure commands and the title.
        A (user-supplied) logo is set too.
\bitem the markup commands are \cs{head}, \cs{sub\-head},
       and \cs{sub\-sub\-head}.\ftn{There is a great variety with respect to
         parameter termination:
         just a parameter within curly braces, or
         |\end...| terminators, or
         enclosed within *-s (TUGboat), or \dots}
        For the markup in the small just plain \TeX{}
        commands---and extensions---can be used.
\smallbreak
\subsubhead{Headline} gets as assignment
\beginverbatim
\headline={\line{\vbox to0pt{\vss
   \centerline{\titlefont\strut\the\title}
   \vskip2ex
   \line{\logo\hfill\vbox{%
     \llap{\strut\the\headtitle}
     \llap{\strut\the\subheadtitle}
     \llap{\strut\the\subsubheadtitle}}}
   \vskip1ex
   \hrule width\hsize height1pt
   }}%end hbox
}%end headline
!endverbatim
\vskip-\baselineskip Similar for the footer.
\subsubhead{Headers.}
Headers for transparencies have a completely different design
goal than for a regular blue script.
I will design \cs{beginhead}  and \cs{endhead},
next to the minimal markup variant |\head{...}|.\ftn{The subhead
   variants are similar but different in detail.}
\subsubhead{The design goals}  are
\bitem a head starts a new transparency
\bitem set the title large (and centered but that is the default for all)
\bitem store the title for use in the running head,
       empty the subhead and subsubhead running head
       variables.
\subsubhead{What are the problems?}
The first head invocation has a different function from
later invocations because it has
to  reactivate \cs{footline} and \cs{headline}.
The pagenumber must be  adjusted to 1 too.
My solution is {\em not\/} a different
coding for head  for the first use,
but to include in the assignation of the \cs{prehead} token
variable a redefinition of itself.
\beginquote
I could not code elegantly the long form and the minimal form
with the argument processed and {\em stored\/} on the fly.
I decided to drop the requirement of processing
the head on the fly for simplicity reasons.
\endquote
\beginverbatim
\prehead{\vfil\eject\egroup\pageno1
 \prehead{\vfil\eject}}
\posthead{\medskip}
\def\beginhead#1\endhead{\headtitle{#1}
 \subheadtitle{}\subsubheadtitle{}
 \the\prehead\centerline{\titlefont#1}
 \the\posthead}
\def\head#1{\beginhead#1\endhead}
!endverbatim
Explanation. The above can best be understood
by realizing that \cs{beginscript}
empties \cs{footline} and \cs{headline}
after opening a group.
The original values for the
\cs{footline} and \cs{headline} are accessible again
on the real first page by closing the group after
the title page has been ejected.
I had to use the endseparator technique,
because the assignation to
a token variable requires an explicit closing brace,
\TB, p.276.
That is the following is {\em not\/} allowed
\beginverbatim
\def\beginhead{...\headtitle\bgroup}
\def\endhead{\egroup...}
!endverbatim
\subsubhead{Bulleted item} is implemented as follows
\beginverbatim
\def\bitem{\bgroup\leftskip\leftskiptrs
  \rightskip0pt plus1fil\relax
  \def\bitem{\item{$\bullet$}}
  \def\smallbreak{\endgraf\egroup
      \smallbreak}
  \let\smallskip\smallbreak
  \bitem}
!endverbatim
Explanation. I chose for to shift the pagecontents
to the right by \cs{leftskiptrs}.
At the title page I could not combine \cs{everypar}
and insert a \cs{bitem}. (infinite recursion.)
\subsubhead{Syntax diagrams} shift to the right as follows
\beginverbatim
\let\bgnsyn\beginsyntax
\def\beginsyntax{\bgroup
 \leftskip\leftskiptrs
 \rightskip0pt plus 1fil\relax\bgnsyn}
\let\endsyn\endsyntax
\def\endsyntax{\endsyn\egroup}
!endverbatim
Explanation. The idea is that some prematerial\Dash to
account for the shifting to the right\Dash is
inserted before the invocation proper.
\head{Related work}
First a disclaimer. It is not possible to really
compare blue.fmt with other formats because of
different goals. What follows below has to be read
as a rough indication of how blue.fmt relates to
the works mentioned.
In a sense \LaTeX{} provides the functionalities I'm after.
However, \LaTeX{} is too complex to master for me, and I don't
like its coding, nor its appearance in print via the standard
styles.
Because of its complexity customization is too time-consuming.
Moreover, it is subject to change and from
the results  I have seen so far
its complexity has increased, alas, and
the end of the change is not in sight.
\LaTeX{} does not worry about adaptations to other
formats because people involved with it consider
\LaTeX{} a (de facto) standard.
The concept of active documents is not in there
apart from what Knuth already provided, albeit under
different names.
Doob's macros for typesetting his `Gentle introduction to \TeX'
are too limited, especially when we consider
the life-cycle of documents.
Berry builds upon plain alone and neglects other achievements.
Spivak provides in his \LAMSTeX{} much of \LaTeX's functionalities
but he refrains from plain's math markup.
The \TeX info.tex macros look as if designed bottom-up.
What is the use of  a meta-parsing macro---very clever there is
no question about that---if we can do without this functionality
in ordinary formats? No attempt for simplicity,
nor a set of common markup tags as a foundation to build upon.
TUGboat's OTR functionality lies at the heart of the page makeup of
blue.fmt. However blue.fmt goes much further than tugboat.sty
in for example the markup for pictures, references, (bordered) tables,
verbatims, and math cross-referencing,
all designed for use with \AllTeX, that is it is not tied up with
\LaTeX{} or plain, exclusively.
Add to that the way how references and pictures are selected
from a database and you will agree that blue.fmt is a leap forward.
The coding of `options' is  less monolithic, and the setup via
modules with thin interfaces is more flexible.
blue.fmt is not a goal per se. It has an open eye for change in general,
and adaptation towards other contexts in particular.
Furthermore, nobody as far as I'm aware of, paid sufficient
attention to how to cope with the balance of
{\em your system\/} in relation with the outer world.
Your own stable format as basis,
adaptable to requirements from the world outside when the
need arises.
\head{Availability}
A problem in combining it all together was how to prevent
redefinition, how to keep all functionalities accessible.
I simply decided to change names, for example
manmac's \cs{output} was changed into \cs{manmacoutput}.\ftn{In analogy
   with \cs{plainoutput}.}
The transparencies macros have been incorporated
as replacement text of the \cs{trans\-pa\-ren\-cies} definition.
blue.fmt is too long to be listed here, some 100K Bytes.
lit.dat comprises  40K Bytes, and pic.dat another 10K.
\subhead{What?}\ \\
Another decision which had to be made is
\beginquote
What is part of blue.fmt\\
and what not?
\endquote
And if something is not included what is the relationship,
can it be used together with blue.fmt?
I decided to lay back and to keep blue.fmt small ;-), but to allow for
extensions. This approach to the contents model of blue.fmt
is sketched in the accompanying diagram.
\topinsert
\null\vskip4ex
\qquad Model of blue.fmt
\bluemodelpic
\endinsert
blue.fmt $\equiv$ manmac + ppt + trs + add-ons\\
add-ons $\equiv$ abbrev. + bib + btable + crossrefs + gkppic\\
\null\phantom{add-ons $\equiv$ } + vrb + databases\\
databases $\equiv$ lit.dat + pic.dat.
A fully automatic index generator for a simple index?
Yes, definitely but that needs some time still.
For the moment use Knuth's IR markup and his formatting of the
enriched IRs, with the step in between via your favourite programming
tool. What about perl?
The file blue.fmt is a composite like a flower.
It starts with an enumeration of the parts included.
Each part starts  and ends with a \%-line,
denoting which part started or ended.
At the end of most, if not all, parts a table of contents
is supplied.
Three comment lines separate the parts.
In principle conflicting names are redefined into local ones.
The databases are not part of the format file because
of selective loading.
For my daily notes I use the markup for the title elements and
the major structure elements from ppt.tex,
with the keywords, abstract and contents markup added.
Also the OTR from ppt.tex is used.
For verbatim, tables, pictures and bibliography handling
I use the macros of the special collections.
Given that those special collections are small---as are
their accompanying  articles---the learning
of how to use these markup tags is not difficult.
It can be argued that blue.fmt contains too much,
especially including manmac looks like an anachronism.
Readers might lose the wood for the trees.\ftn{There is some truth
   in there and we must not confuse the learning of \TeX{} per se
   with becoming familiar with the markup tags needed. The latter can
   be listed on the backside of an envelope.
   However, I consider the exercise and answer macros useful.
   The writing of index reminders and formatting these
   are also things you will need sooner or latter. The writing of
   selected pages to the dvi file is what I do
   occasionally, especially when deprived from dvitodvi.}
I decided to leave manmac in, especially for those who work in
1-column format. Then manmac's specifics are useful.
This also illustrates that switching from 2-column into 1-column
format needs much more extras.
Hopefully, it does not hinder much.
Thinking positively, it allows a growth path.\\
And \dots when you finally will write a book, I'm convinced that
manmac's extras will come in handy. For authors who write in
\AllTeX{} about \AllTeX{} manmac is indispensable.
\subhead{Where?}
Although this article bears the title BLUe.fmt,
the filenames are without uppercase letters: \\
blue.fmt, lit.dat, pic.dat and
the article fmt.art.
\beginquote
the CTAN sites in directory \dots, and\\
NTG's fileserver in directory \dots
\endquote
\head{Looking back}
The hardest thing was not to mix up all the influences I have
been exposed to.
To lay back and to decide about simple coding conventions,
was not easy, especially when the material has not yet been
mastered completely.
When to use token variables and when to use definitions
is also difficult when trying to prevent confusion at the user level.
It is tempting to introduce parameter separators, but I refrained
from those in the outer level macros, and used the straight
forward  parameter mechanism.
After having gained some
practice with the |\this<tag>| idea, I'm still very
pleased with it, especially because of its local character
and no parsing overhead.
\subhead{Summary of tags.}
I have provided in the accompanying table  the major tags with the context
where they are used.
%\topinsert
{\smallskip\small
\begindisplay\displayindent\parindent
\multispan3{\hfil Tags and context of
        implementation\hfil}\cr
\noalign{\vskip3ex}
Tags  &Reg script\quad&Tansparencies\cr
\noalign{\vskip.5ex\hrule\vskip1ex}
\multispan3{Token variables\hfil}\cr
\cs{abstractname}  &+      & \cr
\cs{address}       &+      &+\cr
\cs{author}        &+      &+\cr
\cs{contentsname}  &+      & \cr
\cs{everyverbatim} &+      &+\cr
\cs{everybtable}   &+      &+\cr
\cs{everypicture}  &+      &+\cr
\cs{everyscript}   &+      &+\cr
\cs{everyreferences}&+     & \cr
\cs{footline}      &+      &+\cr
\cs{headline}      &+      &+\cr
\cs{headtitle}     &       &+\cr
\cs{issue}         &+      & \cr
\cs{keywordsname}  &+      & \cr
\cs{netaddress}    &+      &+\cr
\cs{pagenumber}    &+      &+\cr
\cs{postbtable}    &+      &+\cr
\cs{posthead}      &+      &+\cr
\cs{postreferences}&+      & \cr
\cs{postsubhead}   &       &+\cr
\cs{postsubsubhead}&       &+\cr
\cs{postverbatim}  &+      &+\cr
\cs{prebtable}     &+      &+\cr
\cs{prehead}       &+      &+\cr
\cs{prereferences} &+      & \cr
\cs{presubhead}    &+      &+\cr
\cs{presubsubhead} &+      &+\cr
\cs{preverbatim}   &+      &+\cr
\cs{subheadtitle}  &       &+\cr
\cs{subtitle}      &+      &+\cr
\cs{subsubheadtitle}&       &+\cr
\cs{thisverbatim}  &+      &+\cr
\cs{thisbtable}    &+      &+\cr
\cs{thispicture}   &+      &+\cr
\cs{thisscript}    &+      &+\cr
\cs{thisreferences}&+      & \cr
\cs{title}         &+      &+\cr
\noalign{\vskip1ex}
\multispan3{Box variables\hfil}\cr
\cs{abstractbox}   &+      & \cr
\cs{btablebox}     &+      &+\cr
\cs{contentsbox}   &+      &+\cr
\cs{keywordsbox}   &+      & \cr
\noalign{\vskip1ex}
\multispan3{Two-part definitions
     listed by rootname\hfil}\cr
    abstract       &+      & \cr
    btable         &+      &+\cr
    center        &       &+\cr
    contents       &+      &+\cr
    keywords       &+      & \cr
    picture        &+      &+\cr
    pictures       &+      &+\cr
    quote          &+      &+\cr
    references     &+      & \cr
    script         &+      &+\cr
    verbatim       &+      &+\cr
\noalign{\vskip1ex}
\multispan3{Structure definitions\hfil}\cr
\cs{bitem}         &+      &+\cr
\cs{continue}      &       &+\cr
\cs{ftn}           &+      & \cr
\cs{lastscript}    &+      & \cr
\cs{newcol}        &+      & \cr
\cs{newpage}       &+      & \cr
\cs{notlastscript} &+      & \cr
\cs{nxttrs}        &       &+\cr
\cs{(sub)(sub)head}&+      &+\cr
\multispan3{Fonts\hfil}\cr
\cs{authorfont}    &       &+\cr
\cs{headfont}      &+      &+\cr
\cs{subheadfont}   &+      &+\cr
\cs{subsubheadfont}&+      &+\cr
\cs{subtitlefont}  &+      &+\cr
\cs{titlefont}     &+      &+\cr
\enddisplay
\smallskip}
%\endinsert
Note the occurrence of picture and pictures. The latter is used
for the specification of the names.
\head{Looking forward}
What about a letter style and the like?
Similar to \cs{trans\-pa\-ren\-cies} one could supply
\beginquote
|\def\letter{...}|
\endquote
and add this to the blue.fmt.
Another approach is to just input your file with letter macros.
What about a report style instead?
Customize the macros with rootnames
script, keywords, abstract,
contents, and perhaps head and ilks.
Add  your \cs{ls} for formatting the references and supply
this via \cs{thisreferences} as  |\ls<you>|.\ftn{The
   function of \cs{ls} is that it formats a reference with the name
   of the reference as argument to \cs{ls}.}
And what about extending the database of pictures similar to the
database of references?\ftn{Please do so, and let me know.}
Similarly for |\def\book{...}|, which is extremely suited to be
based upon manmac. Earlier, in `Manmac BLUes,' I  discussed
the customization of manmac. For the most likely items
to be customized I provided the following simplifications.\ftn{Also
   provided in blue.fmt. The old beginchapter has been renamed into
   man\-mac\-be\-gin\-chap\-ter.}
%\topinsert
\centerline{Customization Manmac}
\beginlines
\beginverbatim
\font\titlefont=cmssdc10at12pt
\outer\def\beginchapter#1 #2#3. #4\par{%
 \def\chapno{#2#3}\global\exno0 \subsecno0
 \def\hl{\gdef\hl{\issue\hfil\it\rhead}}
 \headline{\hl}
 \def\\{ }\xdef\rhead{#4}
 {\let\\\cr\halign{\line{\titlefont
  \hfil##\hfil}\\#1 #2#3 #4\unskip\\}}
 \bigskip\tenpoint\noindent\ignorespaces}
\footline{\fiverm\rlap{\the\year}\hfil
 --\folio--\hfil\llap{\copyright cgl}}
\outer\def\endchapter{\vfill\eject}
\output{\plainoutput}
\hsize18cm \pagewidth\hsize
\vsize25cm \pageheight\vsize
\issue{MAPS 93.1}
!endverbatim
\endlines
\rightline{Simple isn't?}
%\endinsert
What remains are the handy macros, the suitable environments, the powerful
\cs{exercise}, \cs{answer}, and the simple mark up and writing to a file
of index reminders, the formatting of the index,
next to the  shipping-out of selected pages.
Definitely candidate for addition is an index module which
allows semi-automatic handling of a simple index all within \TeX.
\head{Acknowledgements}
There are too many people to be named whom I owe much while
working on both the BLUes and BLUe's series.
They all have helped me a lot.
Because  manmac.tex, AMS styles and TUGboat styles are all in the
public domain I could learn from these, ponder about alternatives,
and continue development from there.
In particular I like to thank NTG for allowing me to bring out
early versions of my papers in MAPS, and
to thank GUST for having invited me to lecture at GUST 94 about manmac,
and the attendees for having asked for a manmac user guide.
Thank you!
\head{Conclusions}
Macro writing with respect to format or style development
is a special way of programming and to embed this within
the realm of software engineering is a real challenge,
and much needed.
It is still wishful thinking to have a full-fledged simple format
which is easy to use, and of which marked up copy can be easily
transformed into a representation suitable for other circumstances.
blue.fmt comes close, mainly because
I thought more than twice about what Knuth has provided to begin with
before writing something of my own,
and because of my handful of outer markup tags, very few indeed.
Therefore, I conclude
\bitem a variety of tools is needed
\bitem blue.fmt $\equiv$ manmac + ppt + trs + add-ons\\
   add-ons $\equiv$ abbrev. + bib + btable + crossrefs + gkppic\\
   \null\phantom{add-ons $\equiv$ } + vrb + databases\\
   databases $\equiv$ lit.dat + pic.dat
\bitem consistency, portability, longevity,
       flexibility, intelligibility, extensibility,
       are useful and necessary adages when providing a format
\bitem when you consider book formatting $\rightarrow$
       forget about ppt and customize manmac
\bitem keep up with Knuth, be realistic.
\smallbreak\noindent
My added value to manmac and gkppic is
\bitem integration of these within a two-column and
       one-column environment, especially articles and
       transparencies
\bitem separation of specification and paste-up
\bitem |\this<tag>{...}| and |\every<tag>{...}| extensions
\bitem |\pre<tag>{...}| and |\post<tag>{...}| parameterizations
\bitem literature and (new) graphics database
\bitem one-pass bibliography handling
\bitem formatting references in the AMS spirit
\bitem cross-referencing of math, references
\bitem outer and inner markup separation
\bitem a verbatim suite
\bitem a bordered table macro
\bitem transparency macros
\bitem customization outer markup.
\smallbreak\noindent
With respect to coding I added the paradigm\Dash a real pearl to paraphrase
Bentley\Dash to add systematically and  automatically a one-part
macro with the same functionlities on top of a two-part macro.
\smallbreak\noindent
My thesis is that users can benefit from blue.fmt
in typesetting like a craftsman, to achieve the quality of Knuth.
The extra bonus is  stability, and simplicity as well.
The disadavantage is investment in learning, which
by trial\-{-}and\-{-}er\-ror takes a substantial
amount of time.
It has all to do with your attitude,
whether you believe that \TeX{} proper will serve a lifetime, and
whether you like to invest in learning plain and manmac as basis.
The reward is freedom.
You are no longer dependent upon
the gurus for what-\-and-\-when.\ftn{Don't
   believe the argument that plain and manmac don't support
   structured markup. That is not true. Peruse the \TeX book
   script and you will know better.}
Furthermore, I adhere De Vinne's adage
\beginquote
`The last thing to learn is simplicity'
\endquote
\noindent
A user is encouraged to look at the results as provided
in the \TeX book and peruse the markup in the compuscript,\ftn{%
   Available in the public domain, also on NTG's
   4All\TeX{} CD-ROM, and undoubtedly others.}
and then simply practise Knuth's way of markup.
\pasteupreferences
\makesignature
\gdef\makesignature{}
\head{Appendix: Table of contents}
\contents{Introduction
\quad Why?
\quad Notations and definitions
Specifications
Typographical design
Markup language
\quad Conventions control sequences
\qquad The bad news
\qquad The good news
\qquad Options
\qquad Place within context
\qquad Changing the defaults
\qquad Processing more than one
\quad What to start with?
\quad Template
\quad Semantics
\quad Outer markup
\qquad Information part
\qquad Chapters and the like
\qquad Running headers and footers
\qquad Bibliography
\qquad Pictures
\qquad Index
\quad Inner markup
\qquad Special paragraphs
\qquad Quotations
\qquad Exercise and answer
\qquad Math
\qquad Tables
\qquad Pictures
\qquad Verbatim
\qquad Special texts
\quad Handy
\quad Syntax diagram
\quad Manmac's \cs{onecol} specifics
Coding
\quad Conventions
\quad Headers
\qquad Design goals
\qquad What are the problems?
\qquad Take a deap breath
\quad AN-Items
\quad Beginning of the blue script
\quad Selective loading of pictures
\quad One-pass handling of references
Practicalities
\quad How to get rid of overfull hboxes?
\quad How to get rid of underfull boxes?
\quad Use of memory
What not?
Software engineering
\quad Correctness
\quad Robustness
\quad Flexibility
\quad Consistency
\quad Documentation
\quad Documentation?
The world outside
\quad MAPS template
\quad blue$\rightarrow$MAPS
Transparencies
\quad Look-and-feel
\quad Markup
\qquad Template
\qquad Semantics
\quad Coding
\qquad Title transparency
\qquad\quad Contents
\qquad\quad \cs{beginscript}
\qquad Regular transparency
\qquad\quad \cs{headline}
\qquad\quad \cs{head}
\qquad\quad \cs{bitem}
\qquad\quad \cs{beginsyntax}
\qquad\quad \cs{beginverbatim}
Related work
Availability
Looking back
\quad Summary of tags
Looking forward
\quad Manmac customization
Acknowledgements, Conclusion, References
Appendix: Table of contents
Appendix: Survey of |\<prefix><root><postfix>|
Appendix: Readmefi.rst
Appendix: blue.fmt
}%end contents
\pasteupcontents
\newcol
\head{Appendix: Survey of control sequences}
The header denotes the minimal one-part macro, \\
the prefixes pre-, post-, this-, every-, and \\
the postfixes -font, -name and -box.
\begingroup
\catcode`/=13
\def/{\leavevmode\thinspace\hbox{%
 \vrule\vtop{\vbox{\hrule\kern1pt\hbox{%
 \vphantom{\tt e}\thinspace%{\tt#1}%
 \thinspace}}\kern1pt\hrule}\vrule}%
 \thinspace} % control sequence token
\catcode`-13 \def-{$\neg$}
\def\header{min\cs pre\cs post\cs this\cs every \cs font%
            \cs name\cs box}
\def\rowstblst{{abstract}{btable}{center}{contents}{head}{keywords}%
   {picture}{pictures}{quote}%
   {references}{script}{subhead}{subsubhead}{subtitle}%{syntax}
   {title}{{verbatim}}}
\def\btablecaption{Control sequences for reg scripts}
\def\data{%
 +\cs-\cs-\cs-\cs-\cs-\cs+\cs+\rs%abstract
 /\cs+\cs+\cs+\cs+\cs-\cs-\cs+\rs%btable
 +\cs-\cs-\cs-\cs-\cs-\cs-\cs-\rs%center
 +\cs-\cs-\cs-\cs-\cs-\cs+\cs+\rs%contents
 +\cs+\cs+\cs-\cs-\cs+\cs-\cs-\rs%head
 +\cs-\cs-\cs-\cs-\cs-\cs+\cs+\rs%keywords
 -\cs-\cs-\cs+\cs+\cs-\cs-\cs-\rs%picture
 /\cs-\cs-\cs-\cs-\cs-\cs-\cs-\rs%pictures
 +\cs-\cs-\cs-\cs-\cs-\cs-\cs-\rs%quote
 /\cs+\cs+\cs+\cs+\cs-\cs+\cs+\rs%references
 +\cs-\cs-\cs+\cs+\cs-\cs-\cs-\rs%script
 +\cs+\cs+\cs-\cs-\cs+\cs-\cs-\rs%subhead
 +\cs+\cs+\cs-\cs-\cs+\cs-\cs-\rs%subsubhead
 /\cs-\cs-\cs-\cs-\cs+\cs-\cs-\rs%subtitle
 %+\cs-\cs-\cs-\cs-\cs-\cs-\cs-\rs%syntax
 /\cs-\cs-\cs-\cs-\cs+\cs-\cs-\rs%title
 -\cs+\cs+\cs+\cs+\cs-\cs-\cs-\rs%verbatim
{\smallskip\small
\btable\data
\smallskip}
\def\rowstblst{{btable}{center}{contents}{head}
               {picture}{pictures}
               {script}{subhead}{subsubhead}{subtitle}{syntax}
               {title}{{verbatim}}}
\def\btablecaption{Control sequences for transparencies}
\def\data{%
 /\cs+\cs+\cs+\cs+\cs-\cs-\cs+\rs%btable
 +\cs-\cs-\cs-\cs-\cs-\cs-\cs-\rs%center
 +\cs-\cs-\cs-\cs-\cs-\cs+\cs+\rs%contents
 /\cs+\cs+\cs-\cs-\cs+\cs-\cs-\rs%head
 -\cs-\cs-\cs+\cs+\cs-\cs-\cs-\rs%picture
 /\cs-\cs-\cs-\cs-\cs-\cs-\cs-\rs%pictures
 +\cs-\cs-\cs+\cs+\cs-\cs-\cs-\rs%script
 /\cs+\cs+\cs-\cs-\cs+\cs-\cs-\rs%subhead
 /\cs+\cs+\cs-\cs-\cs+\cs-\cs-\rs%subsubhead
 /\cs-\cs-\cs-\cs-\cs+\cs-\cs-\rs%subtitle
 +\cs-\cs-\cs-\cs-\cs-\cs-\cs-\rs%syntax
 /\cs-\cs-\cs-\cs-\cs+\cs-\cs-\rs%title
 -\cs+\cs+\cs+\cs+\cs-\cs-\cs-\rs%verbatim
{\smallskip\small
\btable\data
\smallskip}
\item{-} not implemented
\item{+} implemented
\item{/} implemented without processing on the fly.
\endgroup
\newpage
\onecol
\head{Appendix: Read Me First}
\beginverbatim
README first                                        July 1994
Subject: Note which accompanies blue.fmt and related articles.
The blue.fmt and accompanying articles which explain details
and show examples are free for personal use.
No profit is allowed to make from it.
When the results are used a proper credit
---be it Knuth or the authors of gkpmac,
  or me for the composition and the fill-ins---
is the least a happy publishing user can do.
Best wishes and good luck,   ---Kees---  cgl@rc.service.rug.nl
Contents
--------
The files are: blue.fmt, (format)
               lit.dat, pic.dat. (databases)
A handy auxiliary is pfile to print blue.fmt with appropriate line numbers.
The articles are:
BLUe's Format            (The umbrella article about blue.fmt.)
BLUe's Graphs            (The gkppic macros are explained.)
BLUe's Cross-Refs        (One-pass cross-referencing with math.)
BLUe's Bibliography      (One-pass bibliography handling.)
BLUe's Verbatim          (Handling verbatims in TeX.)
BLUe's Transparencies    (The making of transparencies.)
BLUe's Index             (To appear 1995.)
Also included are articles about macros which are part of blue.fmt:
Manmac BLUes             (Is all about manmac, as-is etc.)
AMS BLUes                (Details with AMS practice, \lsams for
                          formatting references is inspired by AMS.)
TUGboat BLUes            (TUGboat styles are discussed. ppt.tex emerged
                          from this which is at the heart of blue.fmt)
Math into BLUes          (The typesetting of math via TeX is detailed.)
Icons via lfonts         (Common matrix icons via LaTeX's fonts.)
Table Diversions         (The typesetting of tables via TeX is treated.)
Auxiliary articles
Sorting in BLUe          (The basics needed for making a simple index.)
FIFO and LIFO sing the BLUes (FIFO paradigm is treated, fundamental.)
Syntactic Sugar          (Discusses various approaches to macros.)
An overview is supplied in
What is TeX and METAfont all about? (NLUUG, 1993)
The relation with SGML is in SGML(TeX,\dots ). (Cork 1990)
For fun are added---and also show macro writing by example---
Typesetting bridge via TeX
Typesettimng crosswords via TeX
Tower of Hanoi, revisited.
Of course all the articles run with blue.fmt in plain TeX.
Kees van der Laan,  Hunzeweg 57, 9893PB, Garnwerd, The Netherlands.
31 5941 1525, cgl@rc.service.rug.nl.
!endverbatim
%}%end script Minimal form has been used
%            for test purposes.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\newpage
\twocol
\head{Appendix: blue.fmt}
\thisverbatim{\catcode`\!=12
              \catcode`\|=12
              \catcode`\;=0
              \unmc
              \input blue.fmt
{\smallskip\small
\beginverbatim
;endverbatim
\smallskip}
}% end script