A breaking change, by default the `both` mode now counts towards the
first use of a term. This in accordance to the documentation.
(updated in bffe5fdc35fcb67aca188fc0642db5bf23530bcd!)
Added the `:use`/`:spend` and `:nouse`/`:nospend` modifiers, in
accordance with the documentation.
(updated in bffe5fdc35fcb67aca188fc0642db5bf23530bcd!)
The 'wants to be referenced from the index/glossary' counter is now
decoupled from the '(first) usage' counter. This is made consistent with
the documentation (updated in bffe5fdc35fcb67aca188fc0642db5bf23530bcd!).
Furthermore, when the 'supplement' mode is used, the term is now NOT
counted as (first) used (by default). This is now consistent with the
documentation.
Add a mode, two modifiers and change the default usage behaviour of some
modes. This documents the glossy 'API' when using terms with respect to
the usage counter.
Important!
At the moment, this is a description of the intended endresult
(documentation driven development). Next commits will bring the code in
accordance with this desired result.
This distinction between 'mode' vs 'modifier' is already a bit visible
when looking at which modifiers can be combined and what output they
produce. This is just formalizing and categorizing that behaviour.
Also create a comprehensive reference table for all modes and modifiers.
This distinction is made in preparation for more modes & modifiers to
control the usage counter, while keeping everything understandable.
Instead of returning an opaque context element with a fixed list of the
pages, return an array of context elements with each a page
(already unique!).
This is conceptually cleaner. Now it's the themes responsability to join
the array of unique pages however it wants. Most times, this will be a
`.join(", ")`, but could be e.g. `.first()` if the theme only wants to
show the page of the first occurance of this term.
Also update the documentation.
Breaking API change in the themes.
No visual differences (tests remain the same), but there are binary
differences.
This modifier was not orthogonal with other modifiers (for example, it
could not be used together with :noindex)
Added documentation and modified test to reflect that @term[] with an
empty supplement has exactly the same result *and* is more orthogonal.
This partially reverts commit 5957432d5d.
Ensure an empty supplement results in glossy returning empty content,
instead of an empty string. Done by returning #none if no article is
wanted.
When the value #none is returned for the article, this can neatly join
with the other values without a trace. In particular, when `term` is
empty content, this ensures that only #metadata remains, exactly like
in the 'hidden' case.
Effectively, now @term[] is identical to @term:hidden, because no empty
string is inserted. This can be seen in the updated test case, where the
spaces around the blank term are now collapsed to a single space by
Typst, instead of keeping two spaces like before.
A next commit will remove :hidden, since it's not orthogonal w.r.t.
other modifiers, while using a supplement is.
If you want to completely override how a term is displayed, you can pass
content in via the square brackets on the term's reference.
For example, a term @www which might normally display as "worldwide web
(WWW)" or something similar, could instead display as FOOBAR by using
@www[FOOBAR].
There are cases where you might want a term referenced (ie from the
Glossary/Index), but don't want to display it. New modifiers
`hide`||`hidden` do exactly that.
This parameter (to init-glossary()) enables/disables the terms in the
document linking to their place in the glossary. However, this adds a
crazy amount of extra queries which sometimes hinders Typst's ability to
converge on documents.
swaits
BurningKarl