Don't works when the line contains tabs about atom-sublime-select

[bigfive]

What do you mean by fails?
The current behaviour is that is selects the nearest selectable text position (so it would snap to the beginning or end of a hard tab). Is that not what you are experiencing?

[ecanuto]

Well, it depends what you call "selectable". Here, it works perfect with all lines that don't contains "tab" character like on last 3 lines of my sample.

Try it, type 123456 and then , 123456 and last type 123456. Selection on third line don't works as espected.

Please let me know if you can't reproduce behavior then I can try to make a video.

Btw, good job. This extension is what is missing on default Atom instalation!

[efc-awebb]

I use tabs that are 4 positions (spaces) wide. Actual tabs, not tabs-to-spaces.

When I attempt to use column selection on lines that contain tabs, it calculates the correct column positions from which to select, but then applies that as a character position (with a tab counting as a single character).

[chowyuncat]

I think there should be an option always keep a rectangular selection, never selecting anything past the mouse cursor's column, and additionally selecting rows and columns where text doesn't exist.

[bigfive]

Oh I see!
Yeah I'm able to replicate that.

I would actually consider that a bug in the atom API. It doesn't consider tab size when reporting columns, or setting ranges by column number.

I'll have a look through the API and see if there is another way that takes hard-tabs into consideration, otherwise I'll have a look at detecting the number of hard-tabs on each line that is selected and accounting for them when handling column numbers (eek).

[efc-awebb]

I looked into the TextBuffer API methods and didn't find anything that might be useful here besides the methods you're already using. I do have some rough draft logic in place that attempts to calculate the correct placement. I haven't looked into a way to retrieve the user's tab settings, but I suspect that won't be too hard - for now, it's hard-coded to 4-width hard tabs.

It's messy, but it's an improvement over the current behavior in terms of hard tab environments. I'll put up a gist tomorrow and hopefully it'll provide a decent starting point for building out a more robust selection handler.

[bigfive]

Actually I just released a new patch that should fix this.

I was confusing buffer ranges with screen ranges.
As far as I can tell:
buffer column = number of text characters from the left
screen column = number of mono spaced columns from the left

Atom API has a nice bufferRangeForScreenRange method. Hey presto!

[efc-awebb]

Hah! That's exactly what I planned on writing myself to clean up my code. Fantastic.