Bastien <b...@altern.org> writes: > But be prepared for dealing with some stubbornness on my side: > documenting the parser does not mean every issue should be solved > thinking in terms of the parser. Sometimes there should be a > tradeoff between what the parser can parse and what the UI should > offer.
Obviously, it depends on the tradeoff. For example, `org-fill-paragraph' contains one. Here is an excerpt of its docstring: For convenience, when point is at a plain list, an item or a footnote definition, try to fill the first paragraph within. This is perfectly fine because it doesn't go against the parser. It merely acts as if the point was elsewhere in the buffer. But it doesn't pretend that the buffer is different. Another possible kind of tradeoff is commands working on a region. Regions don't mean anything for the parser, so commands acting of them may ignore what the parser knows about the buffer. Nevertheless, in other cases, I highly suggest to never discard the output from the parser. A command ought to always consider that the parser is correct. So, if it means to go against it, it should be specified explicitly somewhere (e.g. C-S-up can clearly break the syntax). Commands activated through simple keys (i.e. TAB) on basic syntax (i.e. an headline) shouldn't, IMO, belong to that category, as it would confuse users even more. Regards, -- Nicolas Goaziou