% language=us runpath=texruns:manuals/musings

% \enableexperiments[fonts.compact]

\startcomponent musings-deserved

\environment musings-style

\startchapter[title={We deserved it}]

\startsection[title={Introduction}]

Among the first macros we cooked up are the ones that we used to typeset
chemistry. They come under the \PPCHTEX, and it was mostly done by Hans, Ton and
Tobias. Over time rendering structures by \PICTEX\ got replaced by \PSTRICKS\ and
later \METAPOST. When we moved to \MKIV\ all that got reimplemented, this time by
Hans and Alan and we stuck to \METAPOST. We actually also updated the syntax and,
because old documents were supposed to be rendered by \MKII\ we took the liberty
to sacrifice some compatibility for a more complete and consistent feature set.
However we never came to updating the existing manual and the new one became work
in progress. We didn't really needed the code and Alan, who did need it, knew
what was there anyway.

Then Mikael had some questions and it pushed me into manual mode: upgrade the old
one and combine it with the pending new one. That also made me wonder if we could
now benefit from some new features in the math engine, which in turn could
simplify the code base. This is typically something that takes weeks instead of
hours so better do it right from the start. And, while at it, one then of course
ends up in looking (again) at arrows (or more specifically: stackers) and from
that to \UNICODE, which in turn is good for some introspection.

\stopsection

\startsection[title={Implementation}]

Among the (old) complications in dealing with chemistry are the following:

\startitemize[packed]
\startitem
    Simple inline chemical formulas (snippets) of the kind \ic { U __92 ^^ 238 ^
    +} where scripts need to be properly vertically aligned and an upright font
    is used.
\stopitem
\startitem
    In running text one wants to use \ic {A + B -> C} but also \ic {A + B <-> C}
    or \ic {A + B <=> C}.
\stopitem
\startitem
    Even more symbolic representations might be on the wish list, think of
    \ic {A - B -- C --- D}.
\stopitem
\stopitemize

There is more but these examples demonstrate a few features: we need prescripts,
proper spacing, line breaks at preferred spots, and some symbols.

\start

\setupalign[profile]
% \showglyphs \showmakeup[boxes] \showstruts
\setupmathstackers[reverse][strut=no]

\startitemize[packed]
\startitem
    And of course one also wants to annotate like \ic {A + B ->{here} C} but also
    \ic {A + B <->{}{there} C} or \ic {A + B <=>{where}{every} C} where the
    arrows stretch with the text.
\stopitem
\startitem
    It would be nice if we can also use the advanced alignment options that are
    available in math but discussing this is beyond this musing.
\stopitem
\stopitemize

\stop

\stopsection

\startsection[title=Unicode]

Implementing the above has never been that hard but just became a little easier
in \LMTX. However, when doing that I wondered if there were more than the already
present symbols to be taken care of. And so I did a search on the internet for
\quotation {unicode and chemistry}. One of the first hits was \quotation {Five
symbols used in chemistry L2/23-193}, a request for some more arrows. One can
search the web for it and see if it is still around. When staring at I wondered a
bit about the descriptions:

\starttyping
BALANCED EQUILIBRIUM ARROW
EQUILIBRIUM ARROW LYING TOWARD THE RIGHT
EQUILIBRIUM ARROW LYING TOWARD THE LEFT
REACTION DOES NOT PROCEED
STANDARD STATE SYMBOL
\stoptyping

When later I discussed this with Mikael we came to the conclusion that \quote
{LYING} probably means \quote {LEANING} but we're not chemists so we can be
wrong. The proposed rendering of the first three boils down to arrows with half a
tip, also known as harpoons: long left and right pointing ones stacked for the
first and long over short ones for the other two.

\startlinecorrection
\setupmathstackers[mathematics][strut=no]
\startcombination[nx=3,ny=1]
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox to 2cm{$\mrightharpoonup {\hskip2cm}{}$}
        \ruledhbox to 2cm{$\mleftharpoondown{\hskip1cm}{}\hss$}
    }} {}
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox to 2cm{$\mrightharpoonup {\hskip2cm}{}$}
        \ruledhbox to 2cm{$\mleftharpoondown{\hskip2cm}{}$}
    }} {}
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox to 2cm{$\mleftharpoonup{\hskip2cm}{}$}
        \ruledhbox to 2cm{\hss$\mrightharpoondown{\hskip1cm}{}$}
    }} {}
\stopcombination
\stoplinecorrection

There are several observation to make here. First of all, there is a whole bunch
of arrows (stacked and single) that bare descriptions mentioning them being
arrows. There is no meaning in them. One can even wonder why some are there. So,
in order to be in sync with that it makes more sense to add a few more harpoons.
Cooking up names like this serves no purpose.

If we look at the proposed shapes (which are actually different from those used
in \TEX\ packages that are references) one can actually wonder about the way
these are supposed to stretch. The short variant is not different from existing
double harpoons in which case the meaning is lost. Then when we go longer we get
this empty space and then we should wonder how the extensible should kick in: how
do the left- and rightmost fixed glyphs and the one or two middle repeated ones
behave?

\startlinecorrection
\setupmathstackers[mathematics][strut=no]
\startcombination[nx=3,ny=1]
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox to 1cm{$\mrightharpoonup {\hskip1cm}{}$}
        \ruledhbox to 1cm{$\mleftharpoondown{\hskip1cm}{}\hss$}
    }} {}
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox to 1cm{$\mrightharpoonup {\hskip1cm}{}$}
        \ruledhbox to 1cm{$\mleftharpoondown{\hskip1cm}{}$}
    }} {}
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox to 1cm{$\mleftharpoonup{\hskip1cm}{}$}
        \ruledhbox to 1cm{\hss$\mrightharpoondown{\hskip1cm}{}$}
    }} {}
\stopcombination
\stoplinecorrection

Why don't we just use the next three? After all, as with math, how we interpret
symbols depends on how we define them to be read. Sometimes Mikael and I get a
good laugh over some of the shapes bound to math code points and we're sure that
some legend is needed when using them.

\startlinecorrection
\setupmathstackers[mathematics][strut=no]
\startcombination[nx=3,ny=1]
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox to 2cm{$\mleftrightharpoons{\hskip2cm}{}$}
    }} {}
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox to 2cm{$\mrightharpoonup {\hskip2cm}{}$}
    }} {}
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox to 2cm{$\mleftharpoondown{\hskip2cm}{}\hss$}
    }} {}
\stopcombination
\stoplinecorrection

Of course there are and can be conventions but getting some agreement is not
trivial. We only have to look at \UNICODE\ math to see some issues, like:

\startitemize[packed]
\startitem
    The repertoire of symbols is large but to a large extent somewhat arbitrary:
    what was known (or assumed) to be used got in there. There is a difference
    between a left, right and even middle bar, just like there is a left and
    right brace.
\stopitem
\startitem
    There are alphabets but they come with holes. We're not supposed to
    distinguish between a Plank constant \quote {h} and variable \quote {h}. We
    can distinguish the greek uppercase \type {A} from a latin \type {A} so
    mathematicians are less convincing.
\stopitem
\startitem
    One can define extensible fences with components but not all of them. There
    are no constructors for arrows. Some likely come from old encodings but
    why add these and not be consistent.
\stopitem
\stopitemize

Plugging a hole in an alphabet is doable. Adding some missing snippets for fences
is also no big deal, just because we have a limited set. Making bars consistent
also is not hard. But I don't see it happen soon. After all, we're decades along
the road and no one bothered much about it till now, and above all, it will never
be complete. I wonder if there ever has been a good analysis, extensive
description and a watertight arguing from the \TEX\ community about what should
go in \UNICODE\ and fonts. It was easy to point to existing fonts and the names
used in macro packages and that shows.

To the above we can add:

\startitemize[packed]
\startitem
    There are some combinations of arrows and e.g.rules missing that could have
    made adding composed constructs easier when there is demand for that. After
    all, this is what existing characters are also used for. And these chemistry
    ones fit into this.
\stopitem
\stopitemize

Our solution is to just accept that the \TEX\ community got what it deserved: a
bit of chaos, unreliable cur'n'paste support, symbols but no meaning, imperfect
\UNICODE\ coverage and therefore imperfect coverage in fonts. The good news is
that we can adapt. But we have to be honest: it is not perfect.

\stopsection

\startsection[title=How about]

As mentioned we have a problem with base glyphs versus extensibles and these
combined hooked things are bad for that. In the chemistry macros the first four
are the ones that we always had and the last two are examples of how we can
render the two leaning extras.

\starttabulate[|c||c||]
\NC \type {<->} \NC \ic {A + B <-> C}\NC \type {<-->} \NC \ic {A + B <--> C} \NC \NR
\NC \type {->}  \NC \ic {A + B  -> C}\NC \type {-->}  \NC \ic {A + B  --> C} \NC \NR
\NC \type {<-}  \NC \ic {A + B <-  C}\NC \type {<--}  \NC \ic {A + B <--  C} \NC \NR
\NC \type {<=>} \NC \ic {A + B <=> C}\NC \type {<==>} \NC \ic {A + B <==> C} \NC \NR
\ML
\NC \type {=>}  \NC \ic {A + B  => C}\NC \type {==>}  \NC \ic {A + B  ==> C} \NC \NR
\NC \type {<=}  \NC \ic {A + B <=  C}\NC \type {<==}  \NC \ic {A + B <==  C} \NC \NR
\stoptabulate

Watch the longer should arrow. Not al fonts have it as extensible, which
indicates that not that much thought has been put in usage patterns. That said,
more interesting would be to use something like this, where we fake a character
from two existing arrows:

\startlinecorrection
\switchtobodyfont[cambria]
\definemathextensible [mathematics] [mleftarrowdashed]  ["21E0]
\definemathextensible [mathematics] [mrightarrowdashed] ["21E2]
\setupmathstackers[mathematics][strut=no]
\startcombination[nx=4,ny=1]
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox{$\mrightarrow     {}{}$}
        \ruledhbox{$\mleftarrowdashed{}{}$}
    }} {}
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox{$\mrightarrow{}{}$}
        \ruledhbox{$\mleftarrow {}{}$}
    }} {}
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox{$\mrightarrowdashed{}{}$}
        \ruledhbox{$\mleftarrow       {}{}$}
    }} {}
    {\vbox{\offinterlineskip\glyphscale2000
        \ruledhbox{$\mrightarrowdashed{}{}$}
        \ruledhbox{$\mleftarrowdashed {}{}$}
    }} {}
%     {\vbox{\offinterlineskip\glyphscale2000
%         \ruledhbox to 2cm{$\mrightarrow     {\hskip2cm}{}$}
%         \ruledhbox to 2cm{$\mleftarrowdashed{\hskip2cm}{}$}
%     }} {}
%     {\vbox{\offinterlineskip\glyphscale2000
%         \ruledhbox to 2cm{$\mrightarrow{\hskip2cm}{}$}
%         \ruledhbox to 2cm{$\mleftarrow {\hskip2cm}{}$}
%     }} {}
%     {\vbox{\offinterlineskip\glyphscale2000
%         \ruledhbox to 2cm{$\mrightarrowdashed{\hskip2cm}{}$}
%         \ruledhbox to 2cm{$\mleftarrow       {\hskip2cm}{}$}
%     }} {}
%     {\vbox{\offinterlineskip\glyphscale2000
%         \ruledhbox to 2cm{$\mrightarrowdashed{\hskip2cm}{}$}
%         \ruledhbox to 2cm{$\mleftarrowdashed {\hskip2cm}{}$}
%     }} {}
\stopcombination
\stoplinecorrection

However, not all math fonts provide these dashed arrows and therefore here we
have to use Cambria (here). And even if it has such dashed hashes, they don't
come with an extensible. I just want to point out that by looking at what we
already have, it might make more sense to extend some (combinations) of those.

It is good to notice that these arrows come with qualifications like \quotation
{code for undetermined script}, so one can wonder how much they relate to math.
Actually the fact that we have these holes in alphabets already indicates that
math is not really seen as script. Occasionally (old) scripts get added and they
get lots of code points, while one can argue for sharing there too, but maybe
their status is higher than the status of math. In \OPENTYPE\ fonts math is seen
as script but that's because it is basically a selector. One can actually argue
for chemistry as a math language in fonts.

The main point I want to make here is that adding some new symbol that is
somewhere used but never made it in \UNICODE\ in the first place needs some
thought. Especially when used in a setting of formulas, where size matters.

\stopsection

\startsection[title=Also]

In \CONTEXT\ we have stackers: text above and|/|or below an extensible, or
extensibles above and|/|or below text. The mentioned arrows are using this
mechanism. However, when playing with \type {\iff}, \typ {\implies} and \typ
{impliedby} in math mode, we noticed some spacing side effects. Originally these
double arrows got skips around them but that interfered with our alignment
mechanism. However, that could be solved without much hassle by letting the
commands check the nature (class) of the previous atom (an indication of being at
the start of a next alignment line). In the end that we decided that an extra
\quote {implication} class was more flexible than adding glue. More interesting
was the observation that in Latin Modern and some other math fonts these arrows
have different dimensions, which leads to yet another alignment issue.

Again that could be solved by taking the usage into account but one can wonder
why the opportunity was lost to make the glyphs consistent with each other, read:
come up with a proper analysis of requirements based on decades of \TEX\ usage.
At least there could have been recommended alternates. But wait, aren't
alternates kind of bad as they demand user intervention (choices)? Sure, but
there are more examples of alternates, take the \quote {\dotlessi} (dotless i).
It has a textual code point but is not in the math alphabets, so one needs
alternates as way out (and yes, fonts then have a blackboard and fraktur dotless
i).

One can argue that these are visual aspects, but with arrows as well as symbols,
we have ended up in a somewhat curious inconsistent situation: there are no
established command names for the single arrows, so there we speak \type
{\..arrow..} while for some there are names, like \type {\iff}. The same is true
for characters like the dotless i and j. Some mathematicians use these in the
same way as some hard- and software vendors put an \quote {i} in front of a
product name, but in order to get it one has to communicate in terms of \type
{\dotless.} or somthing with an \type {i} in the macro name.

\stopsection

\startsection[title=Conclusion]

To come back to updating chemistry. It makes no sense to add much more.
Implementing the left- and right leaning is easy with existing hooks so this is
what we will do. Proper math fonts have these and how likely is it that existing
math fonts get new ones? If it ever comes to more chemistry in \UNICODE\ a
handful of them will not help much. It is probably not that hard to find an
existing symbol that can act as standard state symbol. In fact, some code points
have several additional descriptions so we could just add some to existing one.
It's not like we have gone over the top with doing that for math yet.

\stopsection

\stopchapter

\stopcomponent