Open dannypsnl opened 3 months ago
Agda treats literate programming by ignore source language's semantic and extract only code block, then generates HTML like below
<pre class="Agda"><a id="1" class="Background"># coproduct
<details><summary>Helpers</summary>
</a><a id="51" class="Markup">```agda
</a><a id="59" class="Keyword">module</a> <a id="66" href="coproduct.html" class="Module">coproduct</a> <a id="76" class="Keyword">where</a>
<a id="82" class="Keyword">open</a> <a id="87" class="Keyword">import</a> <a id="94" href="foundation.html" class="Module">foundation</a>
as we can see, the original markdown are put into class="Background"
. I expect the modification I'm doing will lead to similar situation.
At this workflow, we can expect that index.agda
is managed by users themself, so invoke agda --html ...
is not the job of agda-tree
anymore, instead, it simply combine *.lagda.tree
with html/*.html
to get *.tree
.
Since the PR https://github.com/agda/agda/pull/7403 is merged, 2.8.0 will have new literate backend, the new workflow is replacing generated html with forester html namespace syntax.
Once most things are done, back port to let agda ignore non-program part in
*.lagda.tree
.