List of system fonts (R, I, B means roman, italic and bold. Ubuntu probably comes with more fonts, but I couldn't find a list to confirm.):
```txt
Windows
----------
Consolas R RI B BI
Courier R
Courier New R RI B BI
Lucida Console R
Mac
----------
Andale Mono R
Courier R RI B BI
Courier New R RI B BI
Menlo R RI B BI
Monaco R
Ubuntu
----------
Ubuntu Mono R RI B BI
DejaVu Sans Mono R RI B BI
```
```css
font-family: Consolas, "Ubuntu Mono", Menlo, "DejaVu Sans Mono", monospace, monospace;
```
Consolas and Ubuntu are professionally designed fonts ([Lucas de Groot][1] and [Dalton Maag][2]), with true, calligraphic italic, so they go at the top of the stack.
Menlo is [based on DejaVu Sans Mono][3], the only difference being a few tweaked glyphs, so DejaVu serves as a fallback for it.
As for Courier New, other than being unreadably spindly, it's the default monospaced font in all browsers, so there's no need to include it in the stack.
The `monospace, monospace;` declaration is, by now, [a standard hack][4] that overrides some browsers' behaviour of defaulting the `monospace` elements to smaller font size. Without it, any relative font size you apply to them will be calculated from that reduced size (seems to be 13 px in all browsers).
[1]: https://en.wikipedia.org/wiki/Luc(as)_de_Groot
[2]: https://en.wikipedia.org/wiki/Dalton_Maag
[3]: http://www.leancrew.com/all-this/2009/10/the-compleat-menlovera-sans-comparison/
[4]: https://stackoverflow.com/questions/38781089/font-family-monospace-monospace
Hidden code blocks are no longer indented with
one additional space (required for doctests to compile in some cases)
Now the behavior is similar to the rustdoc's
This commit:
- Adds an Option<String> field to the BookConfig which should
contain your google analytics ID
- Allows the google analytics ID to be extracted from the config
file (key is google_analytics_id)
- Adds a test to make sure the field is populated from a config
file correctly
Header fragment links now use "id" attribute instead of the depreciated
"name" attribute.
Similar headers are given numbered ids to avoid id collisions.
For instance, if there are three headers named "Example", their ids
would be "#example", "#example-1", and "#example-2" respectively.
This project already had a transitive dependency on regex; let's use it.
This isn't the most efficient solution, but it should be fine. It ends
up doing five full scans of the text. There's probably an easier way but
I'm mostly just trying to get this to work for now.
This also implements the same algorithm that rustdoc does for generating
the name for the link.
Fixes#204
