[ESS] proposal re: underscore, part 2

[Jack Tanner]

Following up on whether to leave the underscore situation
unchanged, I tallied up the mailing list responses (bottom of
this message). Please pardon me if I misinterpreted anyone's
thoughts.

People seemed to rally around the compromise to create a new
option to bind "<-" insertion to another key. This would satisfy
my original request with respect to editing R buffers, but it
would not help with editing BUGS / JAGS code, and it would not
help ESS novices. If I could rebind smart-underscore and if it
could be disabled for BUGS / JAGS altogether, that would help me,
at least.

Parenthetically, I agree that "<-" is the preferred style of
writing assignment in R, and I use it in my own code. But I
disagree that smart-underscore is a good way of teaching that.

Henning Redestig points out that pasting code containing true
underscores into an ESS R buffer with smart-underscore enabled
will mess up the code. That seems like a serious bug, but it
could perhaps be decoupled from this issue. Henning suggests to
activate smart-underscore only if an underscore is followed by a
space. This change would trigger the pasting bug less often than
the status quo, but it will still snag legal identifiers such
as "foo_".

Here's my tally of the posts on this. Folks that seem to think
that the status quo is problematic in some way:

Vitalie Spinu
Henning Redestig
Chris Evans
Jack Tanner

Folks that are OK with the status quo, but would accept a new
option to bind "<-" insertion to some other key:

Rainer Krug
Richard M. Heiberger
Joshua Wiley
Gerald Jean

Folks that would rather see no changes:

Brendan Halpin
Rodney Sparapani
Tim Hesterberg
Martin Maechler