[TxMt] Re: Elastic Tabstops

[joseph davison]

On Jan 17, 2010, at 2:37 PM, Joshua Kolden wrote:
> If it hasn't already been suggested I think elastic tabstops would be a good feature:
> 
> http://nickgravgaard.com/elastictabstops/
> 


An interesting idea, and it'd be fun to try it out, but in playing with the java demo on that page, I hit a couple of places where the behavior was not what I expected.  Eventually I understood why it was happening, but I predict that it would generate a lot of support calls if there was not a lot of documentation about "quirks".

The basic problem is that a table formatting scheme is very limiting, particularly with automatically detecting table boundaries.   Sometimes "in the middle of a table" you'll want to change the column widths -- essentially, you really want to end the previous table and start a new one.   I think inserting a blank line may accomplish that, but, as I say, it'll take some practice, and there was something strange about how his demo handled blank lines, because it didn't always do what I expected..

I hit the problem in the code.  If you do something like (tabs represented by '>'):

...
> if( some test) { > /* special case */
>> doit_now(x)
>}>/* all done */
...

you have a two column table and are probably happy with the comments, but will be very frustrated when the "doit_now(x)" line, instead of being 'properly' indented, goes into the comment column.

Instead of one table there, you probably want 3, or a 3-column table, with two tabs before the comment.  And, of course it gets nastier if you need doubly indented blocks.

joe