[TxMt] Input Environment Variables

[Charilaos Skiadas]

On Feb 4, 2007, at 9:20 AM, Wolfgang Machert wrote:

> On 04.02.2007, at 14:36, Charilaos Skiadas wrote:
>
>>> Is it possible to have some additional TM_INPUT_END_(LINE|INDEX)  
>>> variables to definitely catch the end-point data no matter what  
>>> kind of selection (downward/upward) has been used?
>>
>> I suppose having TM_SELECTED_TEXT would allow you to determine the  
>> end of the selection. Not ideal, but possible probably.
>
> Although you are right at this point (and the effort doing so is  
> indeed pretty zero - using stdin for larger selections seems like  
> the proper approach) it doesn't seem... appropriate... to me to  
> recalculate some data that is already there - at least in TextMate.

Oh I agree, I also would like a lot more ways of telling TM commands  
what to accept as input and where to output and such, I was just  
offering a possible workaround until the time that such things get  
added. Allan is working full-time on the 2.0 release, so if it's not  
already on his plans for that it might take a while until you see  
this added, even if you manage to build a case for that.

I've thought many times that it might be nice to have a Ruby library  
that does these computations once and then we can all reuse them, but  
I haven't gotten around to writing it yet. I think this is a much  
more manageable idea, that we as bundle developers could implement  
without requiring changes in the core TM program. I am envisioning a  
module that would have a bunch of methods returning line and column  
numbers for all sorts of things, as well as the text before and after  
the current selection, the current scope etc. In other words, a  
module offering even more convenient access to the dynamic  
environment variables and what they allow us to compute.

> Though if I'm the only one having the need for this it might be  
> simpler just calculating the endpoint.
>
>
>>> It would also be convenient having both the start-/end-point line  
>>> content (not only the current) as an environment setting which  
>>> saves reading the whole document for one of two lines again.
>>
>> I would suggest filing these as feature requests with the ticket  
>> system: http://macromates.com/ticket/
>> Make sure to provide practical use-cases that show the need for  
>> these.
>
> My intention is to check the first character left of the selection  
> and the first character right of the selection on alphanumeric-ness.
> AsciiDoc, for which I'm writing a bundle at the moment,  
> differentiates between normal and inline formating (i.e. normal  
> *bold*, or inline **bold**nes) - and I'd like to handle the  
> difference automagically by one single command.

Interesting, I wonder if you could somehow utilize the scope system  
for that, and manage to have a different scope selector for the two  
cases. That would mean changing your grammar to some extend, and I  
don't even know if it would be possible then. Once we get dynamic  
scopes, it might perhaps become possible.

In any case, you only need to read and replace at most the current  
line, which is still much better than what I've had to do in a lot of  
my commands, which is read and replace the entire document ;).

> greetings
> Wolfgang
>

Haris