This is the documentation for HP TeX, version A.00.00, Feb. 14, 1984.
Copyright 1984 Hewlett-Packard Co.
NOTE 1: Plain is read in as a separate file.
NOTE 2: "@" temporarily becomes a letter so that HP TeX can use the
internal control sequences of PLAIN, and to avoid conflict between
the internal control sequences of HP TeX and user names.
NOTE 3: This defines leaders used in the table of contents macros.
<<FONTS>>
NOTE 4: HP TeX has fonts available in sizes from 5 points to 24 points. The
fonts loaded in Plain are listed as comments to make it easier to
verify the fonts available.
NOTE 5: In case a font is used in a write or message.
NOTE 6: \fontdef#1#2 defines the control sequence for the font, but the
font is not loaded until the first time the control sequence is
actually used. The first parameter is the control sequence to be
defined; the second is the external font name.
<<MACROS TO SET FSTYLE>>
NOTE 7: These macros set the family and the background style. Then using
\setfont they set the current font.
<<MACROS TO SET FSIZE>>
NOTE 8: These macros set the background font size, and then use \setfont to
set the current font. \baselineskip and \strutbox are set to
values appropriate to the font size. The fonts in families 0, 1,
and 2 are changed to appropriate sizes. The fonts that are not
already loaded, are not loaded unless they are needed by an entry
into either math mode or display mode.
<<MACROS TO SET CURRENT FONT USING FSTYLE AND FSIZE>>
NOTE 9: \tryfont#1 sets \nofont@ to true if the control sequence determined
by the current background font size and style (#1) has not yet been
defined. If the control sequence does exist, \nofont@ is set to
false, and the control sequence is invoked.
NOTE 10: \setfont uses \tryfont to set the current font based on the
background font size and style. If there is no control sequence
defined for that font, then \tryfont is used again to check for a
font with the background size and in roman style. If there is
still no control sequence defined for that font, the size ten and
style roman is tried. This will always succeed in HP TeX.
\ffam will be a third factor in determining the current font when
other font families become available.
<<DEFAULTS>>
NOTE 11: The default values in HP TeX are family0 and font \tenrm with a
background style of roman, a background size of ten points, and a
superscript/subscript size of seven points.
<<LINE MACROS>>
NOTE 12: These line setting macros take \leftskip and \rightskip into
account.
<<MORE MACROS>>
NOTE 13: This replaces Knuth's definition of \narrower. \narrow@ is used in
the note and warning macros.
NOTE 14: \need is followed by a dimension x. It uses \need@ to determine
if x space remains on the page, and if not, causes an eject to
the next page.
\need@ depends on \okbreak which merely encourages a page break
at that point. Its effect is dependent on its nearness to the
bottom of the page, and whether the text following the \need
fits completely on the page or not.
NOTE 15: \for provides a for-loop. Parameter #1 (\fcount@) contains the
variable name; #2 is the initial value of the variable; #3
(\fexit@) is the terminating value; #4 (\floop@) contains the
instructions to be done in the loop. The loop will not execute
unless the initial value (#2) is less than or equal to the
terminating value (#3). \next@ is set to \fnext@.
NOTE 16: \fnext@ provides the looping capability for \for. It executes
\floop@ and then checks to see if \fcount@ has reached the
terminating value. If it hasn't, then \fcount@ is incremented. If
the terminating value has been reached, \next@ is set to \relax.
\next@ is then entered and depending on its current meaning, either
the loop is repeated via \fnext@, or \relax ends the looping.
NOTE 17: \uline#1 makes an \vbox containing #1 with a horizontal rule .23em
under its baseline, and then creates an \hbox containing the vbox
lowered by the amount necessary to place the baseline of #1 in line
with the baseline of the current line of text.
<<START AND FINISH GROUP STRUCTURE>>
NOTE 18: \start#1 and \finish#1 define local nestable blocks. In the
event of any error, the error message is delayed until corrections
are complete.
NOTE 19: \start#1 uses \ifinvalid@ to determine if #1 is a valid
blockname. If it is, a block (group) is begun and \blockname@
stores #1 as the current block. The control sequence \BEGIN#1 is
invoked and \Error is called to print out any error messages.
NOTE 20: \finish#1 calls \ifinvalid@ to determine if #1 is a valid
blockname. If it is valid, \matchup@ is called to close block #1.
\Error is called to print out any error messages.
NOTE 21: \ifinvalid@#1 tests to see if the control sequence \BEGIN#1
has been defined. If it has not been defined, it is ignored
and \errset is called to store an error message.
NOTE 22: \matchup@#1 closes blocks as follows: if a \finish is used when no
block currently exists, it is ignored. if a finish does not match
the current blockname, blocks are closed until either a match is
found or all blocks are closed. The parameter is the parameter to
\finish.
NOTE 23: \done closes the current block.
<<GENERAL ERROR HANDLING>>
NOTE 24: These macros are used for general purpose error handling.
\errfalse clears the error status. \errset assumes an argument of
an error message; any use of \errset sets the error status and
produces an error message when \Error is encountered.
<<DATE AND TIME MACROS>>
NOTE 25: \mon sets the current month with the current font.
NOTE 26: \date sets the current date: <month> <day>, <year>.
NOTE 27: \hour sets the current time of day in 12 hour format:
<hour>:<minutes> <AM or PM>.
<<INDENTSTYLE>>
NOTE 28: The switch \ifindent acts as a flag to indicate when \indentstyle
is being used. This is used in \indentstyle, \noindentstyle, and
\indentspace.
NOTE 29: \indentstyle sets \parindent to \indentsize. If \indentstyle is
already in effect, \parskip is not changed. If \indentstyle is
not already in effect, \parskip is decreased by 5 points.
NOTE 30: \noindentstyle sets \parindent to zero points. If \indentstyle is
in effect, \parskip is increased by 5 points, otherwise it is not
changed.
NOTE 31: \indentspace will update \parindent only if \indentstyle is in
effect.
<<RAGGED MARGIN>
NOTE 32: \fixmargin@ preserves the original \rightskip.
NOTE 33: \justify undoes the effects of \raggedright by returning to the
original \rightskip, and resetting the interword and intersentence
space to the default values.
NOTE 34: \raggedright saves the original \rightskip with \fixmargin@, adds
some stretch to \rightskip, and if \tt is not in effect, fixes the
interword and intersentence spaces.
NOTE 35: \tt@ is used in \raggedright in the \ifx comparison. It is
necessary if the comparison is to succeed.
<<ITEM AND SUBITEM>>
NOTE 36: \itemset@ takes three parameters: #1 specifies the left indentation
(in multiples of \inset), #2 should be the sum of the left and
right indentation (in multiples of \inset) , and #3 is the tag to be
hung to the left of the item. The line size is in \dimen@. \parshape
is used to build the paragraph. The tag (#3) is hung to the left of
the paragraph.
NOTE 37: \itemlist#1 is syntactically the same as \item, however \itemlist
preserves \leftskip and \rightskip, and both the left and
right margins are indented by the value of \inset. The parameter
contains the tag for the item. Each time \itemlist is used,
the subitem tag is reset, and the item tag is incremented if
appropriate.
NOTE 38: \subitem#1 is syntactically the same as \itemitem, however left and
rightskip are preserved, and both the left and right margins are
indented by twice the value of \inset. The parameter contains the
tag for the subitem. Each time \subitem is used, the subitem tag
is incremented, if appropriate.
NOTE 39: \itempar creates a paragraph identical to that created by \itemlist
except it will have an empty tag. It can be used for the second
paragraph of an item.
NOTE 40: \subitempar creates a paragraph like that created by \subitem, but
with an empty tag.
<<ITEM TAG AND SUBITEM TAG MACROS>>
NOTE 41: The definition of a tag must do three things:
1. define a control sequence \reset[sub]items which will be
used to restart tag allocation between lists;
2. restart tag allocation;
3. define a control sequence \[sub]itemtag to produce a tag.
\icount@ (\scount@) holds the value used to determine the current
item (subitem) tag for numbered and lettered tags.
\idigit@ (\sdigit@) is used with lettered tags to keep track of how
many times the item (subitem) tags have gone through the alphabet
without being reset.
NOTE 42: \itemn@#1 is used in the allocation of numbered tags. The parameter
is either \icount@ or \scount@.
NOTE 43: \itemr@#1 is used in the allocation of lower case Roman tags. The
parameter is either \icount@ or \scount@.
NOTE 44: \itemR@#1 is used in the allocation of upper case Roman tags.
The parameter is either \icount@ or \scount@.
NOTE 45: \iteml@#1#2 is used in the allocation of lettered tags. If the
end of the alphabet is reached, it begins again with "aa" or
"AA". The first parameter is either \icount@ or \scount@. The
second parameter is either \idigit@ or \sdigit@.
NOTE 46: \item@#1,#2,#3 takes three parameters: #1 is the initial value of
\icount@ (item counter), #2 is the value of \idigit@ which is
used to print out the appropriate number of letters when using
letters as tags, and #3 is the definition of the \itemtag. It
defines \resetitems to be the initialization of \icount@ and
\idigit@.
NOTE 47: \itm does not take a parameter. It uses the current \itemtag
as the parameter to \itemlist. If the \itemtag is numbers or
letters, a period is appended to the tag.
NOTE 48: \subitem@ is similar to \item@.
NOTE 49: \sitm is the subitem macro equivalent to \itm for item.
NOTE 50: The defaults are \numbereditems and \letteredsubitems.The
execution of these macros defines \itemtag, \subitemtag,
\resetitems, and \resetsubitems.
<<NOTE AND WARNING MACROS>>
NOTE 51: \BEGINnote defines the blockname "note" and sets the indentation
at twice the left and right skip amounts.
NOTE 52: \BEGINwarning defines the blockname "warning," sets the indentation
to twice the left and rightskip amounts, and places a horizontal
rule above the text. \ENDwarning places a horizontal rule below
the text.
NOTE 53: \marginrule#1 defines an \hrule limited by \leftskip and
\rightskip. Its only parameter specifies the height of the rule.
<<VERBATIM MODE>>
NOTE 54: In verbatim mode, all characters except (\), ({) and (}) are treated
verbatim. Characters that cannot be extracted from the current font
are taken from the math font and given a fixed space equal to the
space of the current font.
<<LUMPLINE AND LUMPSPACE MACROS>>
NOTE 55: These macros make the verbatim mode use less memory. Multiple lines
and spaces are lumped together to make one large glob of glue
rather than many small globs.
The carriage return is made into an active character, and given
the meaning of \lumpline@. \lumpline@ accumulates sequential
carriage returns into one lump of vertical glue in \skip@.
'\ ' is made into an active character, and given the meaning of
\lumpspace@. \lumpspace@ accumulates sequential spaces into
one lump of horizontal glue in \skip@.
<<PAGE CONTROL MACROS>>
NOTE 56: \newpage first balances multi-column format which transfers the
contents of \box 255 to \contrib@, so it always looks like the
page is empty immediately afterwards. The box \contrb@ is checked
to see if the page is really empty.
If \newpage occurs at the bottom of a full page, the \null\vfill
remains in recent contributions and causes an extra page to be
ejected. The workaround is to insert an \eject in the document
just before the \newpage (or the macro containing the \newpage).
The same problem occurs with \evenpage and \oddpage, and
the same workaround is effective.
<<HEADING AND FOOTING MACROS>>
NOTE 57: Headlines and footlines are constructed as follows:
(1) under normal conditions the centered heading or footing is
centered, but in the event that one (or more) of the headings or
footings are too wide, space is distributed if possible so there is
no overlap. The macro triplehead@ does this.
(2) inside and outside headings and footings have priority over
left and right headings and footings. \buildline takes care of
this.
Notice that if the tokens in an \hbox have width 0pt, it is
assumed they are empty.
The macros and token registers associated with headings and
footings should be used near forced page breaks or at the begining
of the document. TeX scans ahead when building a page, so the use
of these macros could affect the previous page.
NOTE 58: \buildline decides which tokens will be on the left and right of
the headline or footline that is being constructed. \lft and \rght
hold the numbers of the boxes containing the tokens to be placed on
the left and right sides of the line. They are initialized to the
boxes for \leftheading and \rightheading. If the box containing the
\insideheading (\box3) is not empty, then for an odd page \lft is
set to 3, and for an even page \rght is set to 3.
If the box containing the \outsideheading (\box4) is not empty,
then for an odd page \rght is set to 4, and for an even page \lft
is set to 4.
\triplehead is called to actually construct the line.
NOTE 59: \triplehead#1#2#3 requires three parameters. #1 is the number of
the box that is to go to the left, #2 is the box number for the
center, and #3 is the box number for the right. \skip@ is the glue
to go between the left and center boxes, and \skip@ii is the glue
to go between the center and right boxes.
If the width of the left box is greater than the width of the right
box, \skip@ii is increased by the difference, otherwise, \skip@ is
increased by the difference.
The line is then built by stringing together the boxes and glue.
NOTE 60: \loadbox#1#2 takes two parameters: #1 is the box number, #2 is the
token list to be put in the box.
NOTE 61: \headline checks to see if headings are turned on or off. If on,
it loads the appropriate boxes with the various heading token
lists and calls \buildline. If off, it adds horizontal glue, and
if \suspendheading is in effect, updates the \headcount.
NOTE 62: \footline is similar to \headline.
<<BOX MACROS>>
NOTE 63: \boxit#1 takes as its parameter the text to be placed in the box,
and can be used when a real box is wanted, e.g., \setbox0\boxit{...}.
\boxit uses \halign to center the lines of text, separated by \cr,
within the box.
The value of \boxline determines the thickness of the rule around
the box. The value of \boxspace determines the amount of space
between the rule and the text.
NOTE 64: \centerbox#1 takes as its parameter the text to be boxed, and then
uses \centerline to center the box resulting from \boxit#1. The box
is separated from surrounding text by \abovedisplayskip and
\belowdisplayskip.
NOTE 65: \textbox#1 takes as its parameter the text to be boxed. It places
the boxed text in the current line (or begins a new paragraph if
encountered in vertical mode) with the baseline of the enclosed
text lined up with the baseline of the line.
<<HP2680A and HP2688A COPY CONTROL MACROS>>
NOTE 66: These commands control the number of copies per page. Owing to the
synchronization problems associated with TeX, these should only be
used near forced page ejects or at the begining of a document.
\globalspecial writes the parameter to the postamble of the
PIF file.
\copycnt contains the number of the register that the PIF server
will look in to find the number of copies desired.
\oldcopy contains the number of copies desired. The default is
two, so if \copieson is invoked before the user has used \copies<n>,
two copies will be generated.
<<HP2680A and HP2688A LOGICAL PAGE CONTROL (LPC) MACROS>>
NOTE 67: \lpcount@ holds the number of the register to be interpreted
by the PIF driver for explicit logical page control.
\nextlp@ holds the page control number to be put into
\count\lpcount@.
These macros avoid the synchronization problems of TeX by
interjecting a page eject at an appropriate time. At the beginning
of the document and after each \shipout, the "empty@" flag is set
to true.
If an LPC macro occurs when the "empty@" flag is true, \nextlp@
is updated, its value is put into \count\lpcount@, and the flag
is set to false.
If an LPC macro occurs when the "empty@" flag is false, \nextlp@
is updated, but the value in \count\lpcount@ is not changed. After
the next shipout, \count\lpcount@ is updated by the output routine.
The codes for \lpreset and \lpexit will, if left alone, only endure
for one page.
<<TABLE MACROS>>
NOTE 68: The table macros are complex. It is recommended that anyone
wanting to understand how they work examine the LOG file of
a small table created with \tracingmacros=1 and \tracingcommands=2.
The table macros use unallocated dimension and token registers to
store the column specifications. New dimensions or tokens should
not be allocated between the \tableformat and the last \tablerow or
\tablebar in a table. If any are allocated the result will be a
very strange looking table and a TeX error. If too many columns,
dimensions, tokens or inserts are allocated, an error message is
given and the registers are used anyway.
A row is processed in sections. \parseline@ takes sections of the
parameter to \tableformat or \tablerow. Each section is delimited
by a ~. \parsecol@ takes sub-sections of the section where each sub-
section is delimited by a |. After a sub-section is processed, the
next sub-section is fetched until the end of a section is reached.
The next section is then fetched and processed until the end of the
original parameter to \tableformat or \tablerow is reached.
\span is temporarily redefined during the execution of either a
\tablerow or \tablebar macro to allow spanning of multiple columns.
The standard table specifications are \leftline, \rightline,
\centerline and (for internal columns) \paragraph. You can also
make your own. There are only two rules:
(1) your macro will have one parameter (the text)
(2) your macro will yield a box of width \hsize.
NOTE 69: \uptoks@ increments \cntA@, used as a pointer to a token register
to hold column alignment specifications, and \cntB@, used as a
pointer to a dimension register used to hold column width specifi-
cations. If too many token registers are used, an error message is
printed, and processing continues.
NOTE 70: \bump@ increments \cntC@, which contains the number of columns
processed in the current \tablerow or \tablebar. If the total number
of columns specified in the \tableformat has not yet been processed,
then \cntC@ is incremented, and \uptoks@ is called to increment the
pointers to the registers that hold the column specifications.
NOTE 71: \tableformat#1 has as its parameter the description of the
table format. The alignment for the entire table is given first,
followed by the formats for each column. A column format consists
of its alignment followed by its width. A | separates the table
format from the column formats, and each column format from the
next.
\tableformat first calls \tablelet with the three parameters
\just@, \just@, and \relax.
Next, \parse@ (which has been assigned the meaning of \parseline@)
is called and sent one parameter, which is delimited by ~. This
effectively sends the contents of the parameter of \tableformat up
to the first ~ to \parse@ (\parseline@). The \done@ is an undefined
control sequence used to mark the end of the specifications in the
\tableformat.
NOTE 72: \tablerow#1 is similar to \tableformat. Its parameter is a list of
the contents for each column of the row. Each column must start
and terminate with either a | for a visible bar, or a ~ for an
invisible bar.
The parameters sent to \tablelet are: \bbar@, \wbar@, and \endrow@.
NOTE 73: \tablebar#1 sets horizontal rules across the table. It resets
\topstrut, \botstrut, and \tablespace, and then passes its
parameter to \tablerow. Its parameter is a list of column
contents, with each column indicated as for \tablerow. A
column can be empty, contain a \vrule, or contain \tbar.
NOTE 74: \tablelet requires three parameters. The first is assigned to the
control sequence \b@, the second to \w@, and the third to \f@.
\CntA@, \CntB@, and \CntC@ are initialized. \parse@ is assigned
the meaning of \parseline@. If currently in horizontal mode,
vertical mode is entered; if in vertical mode, interline glue is
inhibited.
NOTE 75: \just@#1 is never invoked directly. The control sequences \b@
and \w@ are \let to be \just@ in \tablelet when \tablelet is
called from \tableformat. \just@ is invoked by \parsecol@ the
first time \parsecol@ is entered. The parameter to \just@ is the
alignment specification for the table.
\just@ assigns the control sequences \b@ and \w@ the meaning of
\tform@, and initializes \maxcol@ to 0. The control sequence
\tablejust@ is assigned the table alignment specification (#1).
NOTE 76: \tform@#1 passes its parameter, which is the set of specifications
for a column, on to \tform@@ with a ~ deliminating the parameter.
NOTE 77: \tform@@#1#2~ updates the pointers to the registers to be used
for storing column specifications, puts the column alignment (#1)
into the current token register, and the column width (#2) into the
current dimension register. Then \maxcol@ is incremented to count
the column just completed.
NOTE 78: \bbar@#1 begins a row that starts with a visible vrule. The
parameter is empty.
NOTE 79: \wbar@#1 begins a row that starts with an invisible vrule. The
parameter is empty.
NOTE 80: \beginrow@#1 reassigns \b@ and \w@ the meanings of \vbar@ and
\ibar@ respectively. The parameter is empty.
Box0, an \hbox that will contain the row, is begun with a \bgroup,
and the assignments that should hold within the box are made.
NOTE 81: \endrow@ finishes Box 0 and aligns the resulting box according to
\tablejust@.
NOTE 82: \vbar@#1 adds a column entry that terminates with a visible \vrule.
The parameter contains the column entry.
NOTE 83: \ibar@#1 adds a column entry that terminates with an invisible
\rule. The parameter contains the column entry.
NOTE 84: \cs@#1 determines if a column entry (#1) spans multiple columns,
and invokes \span@ with the appropriate parameters.
NOTE 85: \parseline@#1~ is called by \tableformat and \tablerow (and
indirectly by \tablebar). Its parameter is the part (section) of
the parameter to the calling macro terminated by the first
occurrence of ~ . If the parameter is \done@, then processing is
complete, and \parse@ is assigned the current meaning of \f@, which
is \relax if \parseline@ was called from \tableformat, or \endrow@
if \parseline@ was called from \tablerow.
If processing is not done, \p@ss is assigned the meaning of
\parsecol@, and then \p@ss is invoked and its parameter is
that part (sub-section) of the parameter to \parseline@ up to the
first occurrence of |. \done@ is an end of processing indicator.
\parse@ is invoked and, depending on its current assigned meaning,
either terminates processing of this \tableformat or \tablerow
parameter, or takes the next section of the parameter of the
\tableformat or \tablerow terminated by ~ as its parameter.
NOTE 86: \parsecol@#1 is called by \parseline@ and has as its parameter that
part (sub-section) of the parameter to \parseline@ up to the next
occurrence of |. It defines the macro \parsemode@: if we are on
the last column, then the currently assigned value of \w@ (\tform@
or \ibar@, depending on whether a \tableformat or a \tablerow is
being processed) is invoked, and \p@ss becomes \lastp@ss; if we are
not on the last column, then \b@ (which is either \tform@ or
\vbar@) is invoked; in either case, \p@ss is called again and will
either terminate processing or get the next section to be
processed.
NOTE 87: \span@#1#2 sets a box spanning the required number of columns and
containing the text in the parameter. \hsize and \z@ are set to
the width and alignment specifications respectively of the first
column spanned. The pointers are then incremented once for each
column spanned, and the width of each column plus the width of a
vertical bar is added to the \hsize for each column spanned. The
first parameter is the integer following \span, and the second is
the remainder of the column contents.
<<SECTION AND HEADING MACROS>>
NOTE 88: These macros are used for head or section level control.
\basehead is the zero pointer value for the four pointers
to the head counters. The head counters count the number of
current headings at each heading level.
\headlevel keeps track of the level of the current heading.
To add a fifth level, do the following:
(1) Change the "\advance\count10by4" command to
"\advance\count10by5". This allocates
the counters for the level number.
(2) Change "\ifnum\hdlevel<4 " to "\ifnum\hdlevel<5
in the \sethead@ macro.
(3) Allocate a new token register, \fifthlevelhead, and
define a corresponding \headE@
(e.g., \def\headE@{\head5\the\fifthlevelhead}).
(4) Modify the \level macro so that it includes \headE@.
(5) Load \fifthlevelhead with some appropriate tokens.
NOTE 89: \setlevelno#1 has one parameter: the levelnumber to be established.
This can be up to 4 numbers, separated by periods. The level number
and the pointer to the head counters are initialized.
.-1. is appended to the parameter to act as an end of parameter
signal.
NOTE 90: \sethead@#1. has as its parameter one of the numbers in the
parameter to \setlevelno. The first number it receives is put into
the head counter for the first level. The second number goes into
the head counter for the second level, etc. Processing is
completed when -1 is received as a parameter, or when 4 levels have
been done.
NOTE 91: \head#1 is called indirectly by \level via \headA@, or \headB@,
etc. It oversees adjusting the \levelno according to the level of
the new heading. Its parameter is the level of the new
heading.
If the new heading is on a level inferior to the current heading,
\uplevel@ is called, otherwise \headlevel is set to the level of
the new heading and that head counter for that level is incremented.
\head@ is called to update \levelno.
NOTE 92: \uplevel@#1 does what is necessary to add a level that has a higher
number than the current level. #1 is the level of the new heading.
NOTE 93: \head@ defines \levelno.
NOTE 94: These allocate token registers for each level description, and
define a corresponding \headA@ (or \headB@ etc.) that calls \head
with the appropriate parameter, and accesses the tokens for that
headlevel.
NOTE 95: \level#1#2 takes two parameters: #1 is the level of the heading,
and #2 becomes the definition of \title. A case statement is
used to invoke the appropriate choice of \headA@, \headB@, etc.
<<CONTENTS FILE MACROS>>
NOTE 96: "\content@" can now be written with \write
NOTE 97: \ctswrite#1 writes its parameter into a file along with the current
page number. If the file does not already exist, one is created.
When the file is read at a later time, @ will not be a character.
\_ is redefined so that it can be read from the contents file.
\content@ is defined with @ no longer a character so that it can be
read from the contents file.
NOTE 98: \contents adds the table of contents if a contents file was
created.
<<FOOTNOTES AND INSERTIONS>>
NOTE 99: The \vfootnote#1 of PLAIN has been modified to accommodate
multi-column format by the insertion of \realhsize to restore the
original \hsize and by adding 1\baselineskip to remove any glue
from \baselineskip.
NOTE 100: \note generates footnotes with sequential numbers.
NOTE 101: \midinsert has been changed from the definition in PLAIN by
the insertion of \p@gefalse.
NOTE 102: \@ins has been changed from the definition in PLAIN by adding a
return to the normal \hsize and \vsize.
NOTE 103: The \endinsert of PLAIN has been modified. The "room left on page"
calculation has been expanded to handle multi-column format: if
there is room, the columns are balanced and the insertion is
appended. The penalty (first item in the \insert) is set to -100 if
the insertion cannot fit on the current page, this keeps the page
from stretching if the insertion can't fit on the page.
<<COLUMN MACROS>>
NOTE 104: The column macros and output routines are interrelated. Every time
the number of columns changes, a false output routine is invoked.
This routine takes the current page (and splits it up if necessary)
and appends it to the contrib box. The \hsize is then reduced (or
restored) and the \vsize is multiplied (or restored). Then, to
every insert which was not void during the false output routine a
null insert is made. This restores the page goal (in case there are
no bona fide insertions, one insertion is always made, it is called
\hacker@). A new page total is then calculated from the height of
the contrib box. The two macros employed to do this work are
\balance@ and \newtotal@. Under no circumstances should the one be
used without the other.
NOTE 105: \balance invokes the false output routine which has the effect of
balancing the columns if in multi-column mode. The page goal is
then updated by \newtotal@.
NOTE 106: \balance@ invokes the false output routine by setting \ifreal@ to
false, and forcing an eject.
NOTE 107: \newtotal@ restores the page total.
NOTE 108: \multicol@#1 is called by the \BEGIN command for multi-column
format. It takes as its only parameter the number of columns
requested. It provides error checking and ignores the command
if already in multi-column mode. If the command is legal, then
the current page is balanced, and the \hsize and \vsize are
adjusted. The insertion registers are modified, the page total
is adjusted, and stretch is added to \baselineskip.
<<OUTPUT ROUTINES>>
NOTE 109: The output routines are complex. It is recommended that anyone
wanting to understand how they work examine the LOG file of a
short file with \tracingmacros=1 and \tracingcommands=2.
There are two types of output routines: \realoutput@ produces a
\shipout and \fakeoutput@ does not.
The \toks registers associated with each insert are now reserved
(as were \dimen, \skip and \count). Any tokens in these registers
will be destroyed when the \output routine is active.
NOTE 110: \realoutput puts the current page into \out@. If currently in one-
column mode, it goes in directly, otherwise the current page is
first split into columns. Box 255 is then set to \contrb@ with
\old@ appended to it. The glue for insertions, \hsize, \vsize,
\leftskip, and \rightskip are returned to normal for the \shipout.
After the \shipout, the flags and counters are updated for the
logical page macros.
NOTE 111: \fakeoutput takes the current page, balances it if necessary, and
puts it into \out@. \out@ is then appended to \contrb@.
If any column must be shorter due to balancing, it ought to be the
rightmost. The \vsplit primitive, however, will usually make the
rightmost column longer. To avoid this problem, columns are split a
little more than 1/2 or 1/3 down. Experimentation shows that adding
1.16667 baselines to the total height for 2 columns, and adding 2
baselines to the total height for 3 columns affords the greatest
probability of an aesthetic break.
NOTE 112: The "@" is no longer a letter.
