Nicolas Goaziou <m...@nicolasgoaziou.fr> writes:
>> #+begin_example
>> -,#+ATTR_TEXINFO: :table-type vtable :sep , :indic asis
>> +,#+attr_texinfo: :table-type vtable :indic asis :sep ,
>> - foo, bar :: This is the common text for variables foo and bar.
>> #+end_example
>
> This change is not needed. Org markup is capitalized in examples to make
> it stand out. It's not great because 1) it is ugly 2) users may think
> Org syntax is uppercase. But changing that is not the point of this patch.
Sorry about that; I also don't like it when contributors sneak in
cosmetic changes.
>> (defun org-texinfo--normalize-headlines (tree _backend info)
>> "Normalize headlines in TREE.
>>
>> -BACK-END is the symbol specifying back-end used for export.
>> +_BACKEND is the symbol `texinfo'; the back-end used for export.
>
> You don't need to document ignored arguments.
The argument was already documented before I touched this, and
additionally the given name was incorrect. Should I leave this
untouched to avoid making unrelated cosmetic changes or should
I remove this line?
>> +(defun org-texinfo--combine-items (tree _backend info)
>> + "Combine certain description list items.
>> +
>> +_BACKEND is the symbol `texinfo'; the back-end used for export.
>> +INFO is a plist used as a communication channel.
>
> Same as above.
This one is an addition so I am dropping it. (The reason I originally
added it in the above addition is that I saw it being present here.)
>> +If the `:combine' attribute of a description list is non-nil and
>> +an item in that list has no body and is followed by another item,
>> +then remove the first item and prepend its `:tag' to that of the
>> +second item.
>> +
>> +Such an item that absorbed the tags of other items is later
>> +transcoded to one `@item', followed by one or more `@itemx'.
>> +
>> +Return new tree."
>> + (org-element-map tree 'item
>> + (lambda (item)
>> + (let ((plain-list (org-element-property :parent item)))
>> + (when (and (org-not-nil (org-export-read-attribute
>> + :attr_texinfo plain-list :compact))
>> + (eq (org-element-property :type plain-list) 'descriptive))
>> + (let ((next-item
>> + (cadr (memq item (org-element-contents plain-list)))))
>
> I suggest to use `org-export-get-next-element', which is not as
> low-level as this.
Okay changed.
In the next commit, in org-texinfo--massage-key-item, I am also using
the low-level approach. That function isn't passed INFO, so I am unsure
what to do there. Looking at org-export-get-next-element it seems fine
to just pass it nil as INFO.
>> - (items (and tag
>> - (let ((tag (org-export-data tag info)))
>> + (items (org-element-property :itemx item))
>> + (items (cond (items
>> + (mapcar (lambda (tag) (org-export-data tag info))
>> + (nconc items (list tag))))
>> + ((not tag) nil)
>> + (t
>> + (setq tag (org-export-data tag info))
>
> OK, but I think this is a bit convoluted. I wonder if we need
> `org-texinfo--combine-items' at all. I.e., `org-texinfo-item' could
> decide to prefix current tag with @itemx if 1) the item is empty 2) the
> list has a non-nil :compact attribute 3) there's another item following
> it. IOW, we do not need to rely on an :itemx property.
I think so, but its bed time. I'll look into it tomorrow.
Thanks for the review,
Jonas
