On Tue, Jun 04 2013, David Bremner <david@tethera.net> wrote: > Austin Clements <amdragon@MIT.EDU> writes: >> >> Right. I think we should both reference the variable and say what the >> default behavior is (there's no reason not to do both). But isn't >> that what these docstrings used to do? > > Looking at the old docstrings in notmuch-show.el, I agree they basically > implement Tomi's suggestion. While I think copying default values of > variables into docstrings creates some minor maintainability traps > (since we then need to remember to look at all the places a variable is > referenced if we change the default value), I'm willing to revert the > patch if people think the tradeoff of better usability is worth it. Well, revert would be the worst option -- maintainability traps there and no reference to the variable used ;/ So either what we currently have in repository (and merge Mark's similar patches) or have both. In addition to my quick suggestion, what is been seen in notmuch-show.el docstrings this is what `split-string' has: ... If SEPARATORS is non-nil, it should be a regular expression matching text which separates, but is not part of, the substrings. If nil it defaults to `split-string-default-separators', normally "[ \f\t\n\r\v]+", and OMIT-NULLS is forced to t. ... By looking the code, doc hardcoded, defconst split-string-default-separators -- from maintainability point of view those are close to each other... IMO what we currently have is OK, unless SomeOne(tm) provides a neat patch and agrees that maintainability is not really a problem here :D > It is unfortunate emacs doesn't provide a way to expand the current > value of a variable in the help string, but there we are. It probably > wouldn't be as easy to understand as hand crafted text in any case. We could have placeholders in *.el files and tune byte compiler to fill in the docstrings >;) Imagine the added bonus we get by the confusion that causes ! > d Tomi