Community
Participate
Working Groups
Code assist is used to "explore" what is available but in many cases it is used to simply complete what I am typing so that I don't have to type all the characters. In this case it is optimal that the list not appear at all. I want to be able to type "just enough" characters (to narrow the possible completions to just one), hit ctrl-space, and have the completion automatically appear. I think this would require 2 changes from the current behavior 1) An option (on by default) to auto-complete if there is only one completion in the list. This saves having the list popup and a keystroke. Ctrl-z will get rid of the completion if it is incorrect. 2) The lists have to be more "precise" (in order to have a chance of narrowing them down to one completion). The problem is type names. If you include all valid type completions then the lists will always be long. I think that a reasonable approach to this problem is to require that type completions are only included in the list if they match on the case of the first character. Requiring users to use proper case on the first letter if they want type completions is not overly restrictive and I suspect that 98% of the time users are doing this anyway. Restricting it to the first letter means that we will not miss SubView as a completion for Subvi.
It strikes me that options are good things here. If I type "s" and hit ctrl-space, it takes a while to build up the list of things beginning with s or S. If I'm confident I'll get the case right, I'd like to restrict it to those matches which *just* being with s. I'd like to then be able to expand it if I wish, without changing preferences. So, there could be a preference: Default code-completion action: [] Case-insensitive matching, sorted case-insensitively [] Case-insensitive matching, correct case first [] Case-sensitive matching with a hotkey to toggle between which you want to use for the *current* code completion when it's up. I know it's a bit more work, but that would certainly be best for me. I can't instantly think of any downsides to it (other than effort) - anyone? Jon
java core now proposes matches in a relevance order that is case sensitive