problem with comment-start

problem with comment-start

[Werner LEMBERG]

[emacs CVS 2004-08-22]

In plain text mode, setting the variable `comment-start' to, say, ">",
a long line like

  This is a test with comment-start which starts on the >right side<, and which causes strange indentation

looks like this

  This is a test with comment-start which starts on the >right side<,
							>and which
							>causes
							>strange
							>indentation

after pressing M-q.

Can this be improved?  Right now these suggestions come to my mind:

  . If `>' appears in column 12 or greater, don't treat it as
    comment-start.  This ad-hoc value should be customizable.

  . If `>' appears in column 1-11, test the previous and next line
    whether we also have `>' at the same position.

  . Provide a new command within emacs, say, C-M-q, which is the same as
    M-q in text mode, and which ignores `>'.

BTW, there isn't a single remark in the documentation of
`fill-paragraph' that it is influenced by the `comment-start'
variable.


    Werner

Re: problem with comment-start

[Stefan]

> In plain text mode, setting the variable `comment-start' to, say, ">",

If you just set comment-start to ">", you're saying "a > anywhere on a line
starts a comment which ends at the next LF".  Which implies that filling
a line like:

   This is a test with comment-start which starts on the >right side<, and which causes strange indentation

should not result in

   This is a test with comment-start which starts on the >right side<, and
   which causes strange indentation

since that would means that fill-paragraph has moved "which causes strange
indentation" outside of the comment, thus completely changed the semantic
content of the buffer.

I.e. you get what you ask for.  If that's not what you want, then set
comment-start-skip explicitly (rather than let the comment code guess it
based on comment-start) to something like "^> *" so as to make it clear that
a ">" only starts a comment if it's at the beginning of a line.


        Stefan