Re: About prog-prettify-symbols

[Ted Zlatanov]

On Thu, 13 Jun 2013 17:09:00 +0200 Tassilo Horn <address@hidden> wrote: 

TH> Hi all,
TH> I have some issues with `prog-prettify-symbols'.

TH> 1. I'd like to get it working with clojure-mode [fn:1].  In clojure,
TH>    `fn' is what `lambda' is in elisp or common lisp.
...
TH>    So what makes clojure-mode so special that it gets messed up?  One
TH>    difference I can see is that c-mode and emacs-lisp-mode derive from
TH>    prog-mode directly, whereas in clojure-mode there's
TH> ;; For compatibility with Emacs < 24, derive conditionally
TH> (defalias 'clojure-parent-mode
TH>   (if (fboundp 'prog-mode) 'prog-mode 'fundamental-mode))

TH> ;;;###autoload
TH> (define-derived-mode clojure-mode clojure-parent-mode "Clojure"
TH>   ...)

TH>    But then I've tried to make it derive from prog-mode directly, and
TH>    it's still messed up in the same respects.

I don't know how to fix this, sorry.  I looked at the code and can't see
where `clojure-mode' font-locking diverges from the other modes I tested.

TH> 2. Would it be possible to make prettification a bit more flexible?  For
TH>    example, in clojure I'd like to display also (defn ...) as (deƒ ...),
TH>    but currently it seems I can only "replace" a fixed symbol with a
TH>    single character.  I think, it would be better if
TH>    `prog-prettify-symbols-alist' would have the form

TH>      ((REGEX GROUP CHAR) ...)

TH>    so that I could have an alist like

TH>      (("de\\(fn\\)" 1 ?ƒ)
TH>       ("fn" 0 ?ƒ))

TH>    to show (fn ...) as (ƒ ...) and (defn ...) as (deƒ ...).

Juanma requested something similar, a string instead of a single
character.  The string is much simpler than using the match data, you'd
just give "deƒ" as the replacement.  I think mixing match data into the
prettification rule makes it less user-friendly even though it's
certainly more powerful.

Ted