Re: [lmi] InputSequence questions

[Greg Chicares]

On 2010-03-25 15:15Z, Vaclav Slavik wrote:
> On Thu, 2010-03-25 at 02:11 +0000, Greg Chicares wrote:
>> Well, I was making a tacit assumption that the popup editor
>> would edit an InputSequence object's internal representation,
> 
> That was my assumption too -- I'm not particularly looking forward to
> duplicating the parser. Let's revisit this if I/we find out that it's
> impossible to work with (normalized into our preferred form)
> InputSequence.

I should mention that, years ago, Vadim translated this into
something like BNF for use by something like boost::spirit.
It has long been my intention to apply that change the next
time I need to revisit the input-sequence parser, but I admit
that I haven't particularly desired to revisit it because...
well...you see why.

The handwritten parser is more or less cloned from TC++PL-3,
which IIRC cloned its parser from the Dragon Book--in case
you want to modify it. But switching to BNF (or whatever)
would probably be easier, and much better in the long term.

> It's probably clear to everybody, but let me explicitly mention it
> anyway: it won't be possible to preserve original form of the sequence
> after editing it this editor.

Yes. If you save and restore a file today, input sequences
are preserved exactly as entered. But there's no "round trip"
after they're modified in the popup editor: their meaning
needs to be preserved, but the choice among equivalent ways
of expressing the same thing can't be preserved literally.
(It might be possible, but it's not remotely practicable.)