wallet/docs/writing_and_editing_documentation.md

# katex title: >- Writing and Editing Documentation sidebar: true ... # Organization My stuff is currently an enormous pile of disorganized, and thus inaccessible, documents. # pandoc Much documentation is in Pandoc markdown, because easier to write. But html is easier to read, and allows superior control of appearance To convert Pandoc markdown to its final html form, invoke Pandoc by the bash shell file ./mkdoc.sh, which generates html. Directories with markdown files contain a shell script mkdocs.sh which compiles them into html, and also executes the mkdocs.sh of each of its subdirectories, if they have a mkdocs.sh The directories also contain a file called navbar, which is the horizontal navigation menu. Each markdown files should start with some yaml data that sets the title. Thus the markdown for this web page starts with ```yaml

katex

title: >- Writing and Editing Documentation sidebar: true ...

- `# katex` tells the bash script and pandoc to support katex
- `sidebar: true` causes the table of contents to appear at the side
- `notmine: true` suppresses the copyright license.

In the windows 10 environment, shell scripts used  in this project need to be
associated with [Git Bash](libraries/git_bash_undocumented.html) or run from within Git Bash.

This vast pile of notes is out of control, and writing code and maths in
html leads to intolerable overheads.

Hence markdown, the popular markdown conversion program
being the open source Pandoc.

Markdown converters are apt to throw a flood of incomprehensible html code
into your final document, taking low level html control away from the writer.

Pandoc, however, will allow you to take control.  To integrate html and css
with markdown using Pandoc is a bit like rolling marbles with your
elbows through a cage. One has to work through and around the entry
points that Pandoc gives you, while if you were writing in html you could
just write what you wanted directly, but having done the work,
Pandoc can then ensure it is done for every document in the same style in
the same way, and you can change the final form of every document in the
same way all at once.

Sphinx is very popular and widely used, and written in the far more
accessible language python, but to access the power of html, css, and
JavaScript one must write a Sphinx theme, and the creation of a Sphinx theme
is less than well documented and appears to be subject to change.

Visual Studio Code theoretically does automatic generation of the html
equivalents of markdown files, but I never was able to get it working
satisfactorily.

Pandoc has a number of powerful extensions that allow integration of html and
markdown, among them markdown native mode divs `:::`

```markdown
::: {style="…"}
	…
:::

And native mode spans […]{style="…"}

Which extensions do not work correctly in Visual Studio Code.

These can be used to put an anchor in text, but the easiest and most intelligible way to insert an anchor is as a header.

Since markdown has no concept of a title, Pandoc expects to find the title in a yaml inline, which is most conveniently put at the top, which renders it somewhat legible as a title.

Thus the markdown version of this document starts with:

---
title: >-
	Writing and Editing Documentation
# katex
...

Converting html source to markdown source

In bash

fn=foobar
git mv $fn.html $fn.md && cp $fn.md $fn.html && pandoc -s --to markdown-smart --eol=lf --wrap=preserve --verbose -o $fn.md $fn.html

Math expressions and katex

Pandoc can render most maths markdown without needing katex, for example:

{e}^{i\pi}+1=0

a=b+c

f(x) = x^2

\sin(\pi/6) = 0.5

\int_a^b f(x) dx

\int_a^b \tan(x) dx

\int \sin(x) dx = \cos(x)

\sum a_i

\lfloor{(x+5)÷6}\rfloor = \lceil{(x÷6}\rceil

\lfloor{(x+5)/6}\rfloor = \lceil{(x/6}\rceil

0\le S \lt 5

Use \bigcirc, not capital O for Omicron \bigcirc. \Omicron will not always compile correctly, but \ln and \log is more likely to compile correctly than ln and log, which it tends to render as symbols multiplied, rather than one symbol.

\ln(1+x)=x-\bigcirc(x^2)

H(a|b|v)

Some maths expressions will break Pandoc unless one tells it to use katex. Stick the # katex header in when something breaks. Pandoc avoids using katex unless it has to.

But for these expressions, it has to:

\sin(\frac{\pi}{6}) = \frac12

\displaystyle\frac{u(x)}{v(x)}

Inline equation \displaystyle\sum\limits_{i=1}^n i^2 = \frac{n(n+1)(2n+1)}{6} text after inline equation

\displaystyle\sum\limits_{i} i^2 = \frac{i(i+1)(2i+1)}6

The square root of 100 is \sqrt{100}=10.
The cubic root of 64 is \sqrt[3]{64}=4

\bigg\lfloor\frac{x+5)}{6}\bigg\rfloor = \bigg\lceil{\frac{x}{6}}\bigg\rceil

A file that does not need katex has the header:

---
title: >-
	Document title
...

But if it does need katex, it has the header

---
title: >-
 # katex
	Document title
...

So that the bash script file ./mkdoc.sh will tell Pandoc to find the katex scripts.

For it offends me to put unnecessary fat in html files.

overly clever katex tricks

spacing control

k \approx \frac{m\,l\!n(2)}{n}%uses\, to increase spacing, uses \! to merge letters, uses % for comments 

k \approx\frac{m\>\ln(2)}{n}%uses\> for a marginally larger increase in spacing and uses \ln, the escape for the well known function ln 

size control

 \exp\bigg(\frac{a+bt}{x}\bigg)=\huge e^{\bigg(\frac{a+bt}{x}\bigg)}%use the escape for well known functions, use text size sets

text within maths

k\text{, the number of hashes,} \approx \frac{m\ln(2)}{n}% \text{} for render as text

katex macro used recursively

\def\mydef#1{\frac{#1}{1+#1}} \mydef{\mydef{\mydef{\mydef{y}}}}%katex macro 

Tables

Pipe table with header and alignment control

Without counting spaces, but without multiline

Pipe table:

Right	Left	Default	Centre
12	12	12	12
123	123	123	the quick brown fox jumped over the lazy dog
1	1	Carrian Corporation	1

And, with less mucking about, alignments

with alignment, without counting spaces, but without multiline

fruit	price
apple	2.05
pear	1.37
orange	3.09

multiline without bothering with pipes

Counting spaces to align. Only editable in fixed font

This allows multiline, but visual studio code does not like it. Visual Studio Code only supports tables that can be intelligibly laid out in visual studio code.

---

Centered Default Right Left Header Aligned Aligned Aligned

---

First row 12.0 Example of a row that spans multiple lines.

Second row 5.0 Here's another one. Note the blank line between rows.

The header may be omitted in multiline tables as well as simple tables

Notice the alignment is controlled by the first item in a column

In this table, edited in a fixed font, you are using whitespace and blank lines to lay out the table. It is unintellible in a variable width font.

---

First row 12.0 Example of a row that spans multiple lines.

Second row 5.0 Here's another one. Note the blank line between rows.

---

Grid tables

Allows multiline, and alignment, but visual studio does not like it, and you still have to count those spaces. Works fine if you are editing in a monospaced font, otherwise unusable

+---------------+---------------+--------------------+ | Fruit | Price | Advantages | +===============+==============:+====================+ | Bananas | $1.34 | Mary had a little | | | | lamb and every | | | | where that mary | | | | bright color | +---------------+---------------+--------------------+ | Oranges | $2.10 | - cures scurvy | | | | - tasty | +---------------+---------------+--------------------+ | Durian | $22.10 | - king of fruits | +---------------+---------------+--------------------+

Alignments can be specified as with pipe tables, by putting colons at the boundaries of the separator line after the header.

+------------+---------+---------------------+ | Left | Right | Centered | +:===========+========:+:===================:+ | Bananas | $1.34 | - built-in wrapper | | | | - bright color | +------------+---------+---------------------+ | Durian | $22.10 | - king of fruits | +------------+---------+---------------------+

For headerless tables, the colons go on the top line instead:

+--------------:+:--------------+:------------------:+ | Right | Left | Centered | +---------------+---------------+--------------------+

footnotes

Markdown permits footnotes.

This is a useless feature present because people are moving legacy documents to the web.

¹ They can be inline ^[this is an inline footnote] but an inline footnote appears at the bottom of the document, which is confusing in html unless the document is short.

Footnotes were designed to be used with paginated^[footnotes are an ancient hangover from paginated documents in order to do hypertext on paper] documents.
Html is designed for hypertext.

Images

Images are preferably *.webp, and are expressed as the markdown code

![](./image_name.webp){style="width: 479px; height: 386px;"}

assuming the actual image size is 479 pixels wide and 386 pixels high, 479 pixels being the typical width of an image rendered at 100% of the text width with my default formats, and indeed a whole lot of other people's default formats on a typical screen. But obviously pixel size is getting smaller, so for device independence and simplicity, might well give a larger image, and leave out the style command.

Diagrams

The best way to do diagrams is svg and the Visual Studio Code scalable vector graphics extensions.

If you place the data directly in markdown, rather than in an svg file, the markdown is a lot more tolerant of errors than the browser when it includes an svg file, so it is very easy to create svg that displays fine, until you separate it into an svg file to be included as an ordinary image file, whereupon it fails to display in the browser.

Error checking in the markdown and in visual studio code svg editors fails to pick up all errors that will prevent an svg file from displaying.

Inkscape files are unreadable, and once they are cleaned up, Inkscape cannot read them. To (irreversibly) clean up an Inkscape file, minify it in Visual Studio Code to get rid of all confusing mystery cruft inserted by Inkscape, edit it back into markdown compatible form, and reinsert it in the markdown file.

A sequence of straight lines is M point, L point, L point.

Z and z draw a straight line back to the beginning, use in conjunction with fill="#red", for example fill="#FF0000". If the line is open, fill="none"

H and h draw horizontal lines, V and v vertical lines. To see which line you are working on, convenient to temporarily begin and end it with a cross, h 3 h-6 h3 v3 v-6 v3

Drawing smooth curves by typing in text is painful and slow, but so is drawing them in Inkscape. Inkscape is apt to do a series of C beziers with sharp corners between them, and when I try to fix the sharp corners, the bezier goes weird.

If you can get your desired curve with a single M point c point point point that causes the least grief, and gives you a nice smooth curve.

The first point is the starting position, the last point is the end position. The direction of the first control point sets the starting direction, the direction of the second control point sets the end direction, and how far the control points are out controls how and where the curve changes direction. If the curve is weird and pathological, there is something funny with your control points.

Some control point positions lead to singularities in the derivative of the curv but for reasonable control point positions, you get a nice smooth curve.

Capital letters indicate absolute points, lower case relative points.

Before deciding you need to add additonal points to get the curve you want, see if you are doing something dumb with your control points. You usually are.

If you want a sharp turn coming out, and a smooth curve, then a sharp curve coming in to the destination, you do not put in a bunch of intermediate points, you just put your control points close to the beginning and end.

The further the control point is away from the end of the previous S curve, the further its influence propagates into the next S curve. Which can have surprising results if the next S curve is very short, so that influence of the previous control point propagates far beyond its end.

If you want a smooth and gentle curve, you want the the reflection of the previous C or S control around the middle of the next S curve

You only need additional points when you want the curve to go through a narrow pass, in which case you are going to have a C curve going to the narrow pass, the last control point before the narrow pass, its reflection in the next S curve after the narrow pass.

When you want to join two points, and don't care about the path, use an L straight line

When you want to join two points, and you care about the direction in which it starts, or the direction in which it finishes, but not both, use a Q, which gives you one degree of control freedom.

When you want to join two points, and care about the direction it starts, and the direction it ends, use a C, which gives you two degrees of control freedom.

When you want to join two points, and care about the direction it starts, the direction it ends, and you want it to go through a gateway in the middle, use a C S, which gives you three degrees of control freedom. But watch out for the reflection of the last control point in the C landing somewhere difficult inside the S curve. If the last control point in the C is further from the end point of the C than the end point of the S, things can get strange. Sometimes you want to adjust the behavior of the S curve by moving the last control point of the previous C or S to the right position. A distant previous control point is apt to have a big effect on the following S, a near control point little effect, but likely to give you an unpleasantly sharp turn at the join.

M point c point point point s point point
s point point ... s point

Is guaranteed to give you a smooth curve, for reasonably sane control points, with the curve passing through the second point of the s, and its direction set by the first point of the s.

You change a control point, the effect is entirely local, does not propagate up and down the line.

If, however, you have a long move and a short move, your implied control point is likely to be in a pathological location, so the last control point of the long curve needs to be close to the starting point of the following short move.

M point q point point t point t point ... t point

<div style="width: 100%; height: auto; overflow: auto;">
<svg
	xmlns="http://www.w3.org/2000/svg"
	width="width: 100%"  height="100%"
	viewBox="0 0 120 80"
	style="background-color:ivory">
	<g stroke-width="2">
		<path fill="none" stroke="#00f000"
			d="
				M14,45,  c40,-20, 30,-56 54,-18,  s55,15 40,15
					 s-15,10 -30,15
				M 5,45,  q 10,20 30,1, t 10,10, t 10,12
			" />
	</g>
</svg>
</div>

Is also guaranteed to give you a nice smooth curve for any reasonably sane choice of the initial control point and the position of the t points, but you cannot easily control the direction the curve takes through the points. Changing the control point of the first q will result in things snaking all down the line, and changing any of the intermediate t points will change the the direction the curve takes through all subsequent t points, sometimes pushing the curve into pathological territory where bezier curves give unexpected and nasty results. Works OK if all your t curves are of approximately similar length.

Scalable vector graphics are dimensionless, and the <svg> tag's height, width, and ViewBox properties translate the dimensionless quantities into pixels. The graphics default to fixed aspect ratio, and anything outside the viewbox is not drawn. To adjust your image's position within the viewbox, you put everything into a single big group, and apply a translate to that group.

The enormous advantage of scalable vector graphics is that it handles repetitious items in diagrams beautifully, because you can define an item by reference to another item, thus very large hierarchical structure can be defined by very small source code.

<div style="width: 100%; height: 22em; overflow: auto;">
<svg
	xmlns="http://www.w3.org/2000/svg"
	width="29em"  height="12em"
	viewBox="40 60 60 50"
	style="background-color:ivory">
	<g  id="startblocks"
		font-family="'Times New Roman'" font-size="5"
		font-weight="400"
		stroke-width="2"
		style="text-decoration:underline; cursor:pointer;" >
		<line x1="22" y1="70" x2="28" y2="100" stroke="lightgrey"/>
		<rect style="fill:#FFFF00;"
			x="12" y="64" width="36" height="20">
			<animate attributeType="XML" attributeName="y"
				from="64" to="120"
				dur="5s" repeatCount="2" restart="always" />
		</rect>
		<text style="fill:blue;"  x="14" y="74">
			<animate attributeType="XML" attributeName="y"
				from="74" to="130"
				dur="5s" repeatCount="2" restart="always" />
			start animation
		</text>
	</g>
	<rect x="60" y="64" width="20" height="20">
		<animate attributeType="XML" attributeName="y"
			from="64" to="120"
			dur="5s" repeatCount="3" restart="whenNotActive"/>
	</rect>
	<g
		font-family="'Times New Roman'" font-size="5"
		font-weight="400"
		stroke-width="2">
		<path fill="none" stroke="#00f000"
			d="M14 101,  c40 -20, 30 -56, 54 -18 s60 15, 40 15  c -20,0  -10,-20  0,-20 s 9,7 8,12" />
		<ellipse cx="60" cy="85" rx="12" ry="5" style="fill:red" />
		<text  x="60" y="82" text-anchor="middle"  style="fill:#A050C0;" >
			A simple scalable vector graphic
			<tspan x="60" dy="8">
				directly embedded in markdown.
			</tspan>
		</text>
	</g>
</svg>
</div>

<script>
	document.getElementById("startblocks").addEventListener
	(
		"click", evt =>
		{
			document.querySelectorAll("animate").forEach
			(
				element =>
				{
					element.beginElement();
				}
			);
		}
	);
</script>
</div>

---

1. This is an out of line footnote, that can be inserted anywhere, which is confusing in an html document, and hard to edit, because they always wind up sequentially numbered at the bottom of the document.

   Reflect on how unusable wikipedia foot notes are. ↩︎