Ignorecase functionality

[dNitro]

CamelCase is not so simple to type. Sometimes fingers don't hold Shift key properly so as soon as you type an uppercase in place of a lower one you missed the tern suggestions becase they are case sensitive. Added case sensitivity as an option so any one have an option to active it or not.

[marijnh]

This seems to only adjust the editor-side code. Shouldn't you also be passing ignoreCase to the Tern query to make it work?

[dNitro]

@marijnh , Thats the reason why i request it for ternforvim. vim completion algorithm is capable of cope with case sesitivity in this way. I think passing ignoreCase shouldn't be needed as i am ok now with this method.

[marijnh]

It seems to me like Tern will filter out relevant completions if you don't tell it to match in a case-insensitive way.

[dNitro]

@marijnh , Oh, understood what you mean. just now tested a word in middle and doesn't get suggestions. so where we should start? Is it trivial like the editor side or hard to implement? I think its a good option to have.

[marijnh]

It's easy, just add an ignoreCase: True parameter to the JSON query object that's being sent over (line 264, on the current master branch)

[dNitro]

@marijnh , cant get it working. could you add it yourself? and still doesn't tell me what is your opinion about this feature. would you consider push it over?

[dNitro]

@marijnh , Actually the parameter is caseInsensitive: True BUT also with this one for example i cant get both Document and document after pressing <C-x><C-o> in the following position: doc| at the same time! (just get document)

[marijnh]

The server's caseInsensitive feature seems to work, when I test it. Could it be that the client is doing its own filtering?

[dNitro]

@marijnh , Yes, its because of client caching! Currently we call the server on non-alphanumeric characters through ".*\\W" regex matching. so if we change it to ".*\\w" it should work and ... yessss. It works :)

[marijnh]

Great! Could you update your pull request so that I can merge it?

[dNitro]

@marijnh, Here it is. Of course i am not sure about ".*(?:\\w|\\W)" pattern! \w and \W are complementary and probably could be exchange with a much nicer one.

[marijnh]

That regexp is definitely dodgy (\w|\W matches everything), and I'm not sure why you need to adjust that line at all -- it test if a non-word character was typed into the line after the cached completion, and discards the cached completions if that's the case. \W simply means the inverse of \w, which matches a word character, and doesn't have anything to do with case.

[dNitro]

@marijnh , If we does not discard cached completions on word charachters typed after the cached completion we does not get case insensitivity at all. so first i exchange ".*\\W" with ".*\\w". and then we also does need to discard cached completions if a non-word charachter types. so i came up with aforementioned pattern for ignorecase option on and also wanted to keep the non-ignorecase logic same as before. So i think replacing the ".*(?:\\w|\\W)" pattern with "." will do the job. Got you any other suggestion? ( I tested with "." pattern and its working great)

[marijnh]

If we does not discard cached completions on word charachters typed after the cached completion we does not get case insensitivity at all

Why? If the server consistently returns case-insensitive results, I don't see why caching is a problem.

[dNitro]

@marijnh, With ".*\\W" pattern in place thats already present in master branch and requesting with caseInsensitive: True query:

i -> docu -> <C-x><C-o>

<C-e> -> m -> <C-x><C-o>

so at first we get case insensitivity but if we close popupmenu and type a word charachter we dont get them anymore. so it seems word charachters somehow interrupt server consistency to return case-insensitive results.

now if we change pattern to "." and requesting with caseInsensitive: True query:

i -> docu -> <C-x><C-o>

<C-e> -> m -> <C-x><C-o>

The latter is so persistent and on every charachter gives the desired case-insensitive result. I dont know how much impact it has on the work that tern server should do, but it is now i use it for one week and dont counter any degradation on terns work.

[marijnh]

(I'm really not going to accept a patch that intentionally sabotages a regexp as a way to disable a block of code.)

[dNitro]

Okay, I very much appreciate that you, @marijnh, took the time to review this patch. Anyway tern is a great piece of software, javascript community is owe you. keep up the good work.