Autocomplete toggle?

[Monte Goulding]

> On 27 Sep 2017, at 3:16 pm, J. Landman Gay via use-livecode <use-livecode at lists.runrev.com> wrote:
> 
> On 9/26/17 8:45 PM, Monte Goulding via use-livecode wrote:
>>> On 27 Sep 2017, at 9:12 am, J. Landman Gay via use-livecode <use-livecode at lists.runrev.com> wrote:
>>> 
>>> Apologies Monte, I'm sure autocomplete will be handy at some point, but is there a command I can issue in the message box to turn it off for a while?
>> We have added a menu Edit > Options > Autocomplete to toggle it. It’s in 9.0 DP 9 and will be in 8.2 DP 2.
> 
> Fantastic, thank you. I don't see that in the release notes.

Look for 20386

>>> My Android phone was cabled to the Mac and when asked, I clicked "yes" to start a remote debugging session. In this case none of my breakpoints triggered (not the fault of remote debugging, my breakpoints were in places that didn't run) so I quit the app. LC never indicated I had disconnected.
>> Hmm… it should do that when the socket is closed as the app is shutting down…
> 
> I remember it used to, back before it stopped asking me at all. I wonder if it's due to the lack of messages sent with the hardware buttons. There's no notice of any kind for the Home and App Switcher keys. And the Back key is iffy.
> 
> http://quality.livecode.com/show_bug.cgi?id=19420
> http://quality.livecode.com/show_bug.cgi?id=20456 <http://quality.livecode.com/show_bug.cgi?id=20456>

Hmm…I'm not sure. It needs investigation.
> 
>>> 
>>> I began to edit a script and the phone went to sleep. There was severe lag, the script editor started missing keystrokes and would then add them to some other place in the line. Every attempted correction caused other characters to be inserted somewhere else.
>> Wow that doesn’t sound good. I hope we can get to the bottom of it so it doesn’t occur again. Are you on 8.2 or 9.0?
> 
> This was in 9.0 dp 9.

OK

> 
>>> I tried to add a comment to a line and autocomplete kept inserting text for me though I didn't select any of the suggestions
>> What kind of text? The same as you are entering but added elsewhere too or something else?
> 
> Here's the whole line. I first added the two hyphens at the front, then tried to add the trailing hyphenated comment at the end which is where it went off the rails:
> 
> -- pass scrollerDidScroll -- jg: don't pass, it blocks scrolling when there are nested behaviors
> 
> It simply wouldn't accept my comment; I ended up typing it in BBEdit and pasting it into the script. If I remember right, it didn't like anything after my initials with the colon. It would miss a keystroke, I'd keep typing, and then notice it had placed a character in the middle of a word a couple of words back. I'd delete the extra character, set the insertion point to where I was typing originally, type the letter again, and it would again insert it back somewhere else. Sometimes it missed keystrokes entirely and they never did show up. Meanwhile the suggestion list was popping up and down repeatedly though I didn't interact with it.
> 
> The lag was severe but intermittent, maybe 2 seconds or so before a keystroke appeared in the script text. That's why I suspected something was going on with the debugging socket. The script I was editing was short, only about 60 lines, so it didn't seem like a delay due to text chunking.
> 
> It would put up suggestions when I wanted to type "don't" and then add the "t" a couple of words back. It wasn't possible to type "scrolling". I never did make it to the word "behaviors”,

It really sounds like things were going haywire. Sorry about that! It’s feasible there were execution errors on IDE stacks that were obscured by gRevDevelopment false.

> I gave up after a few minutes and typed it in BBEdit.

I can see why!

> 
>>> (shouldn't it ignore anything to the right of a comment delimiter?)
>> It should probably only present completions that are suitable within a comment context. I’ll need to think about that a bit.
> 
> 
> I'd encourage ignoring comments. It's similar to a problem we've had for ages when I want to add a comment like this:
> 
>  doSomething  -- else the world ends
> 
> I have to change those to:
>  doSomething -- otherwise the world ends
> 
> 
>>> Eventually the script got so mangled I just quit without saving. I think if I had been able to turn off autocomplete from the message box I could have corrected and saved my work.
>> What this sounds like (at least I think it sounds like) is an autocomplete placeholder got messed up somehow causing typing in one place to also add the text elsewhere. If that’s the case then turning autocomplete off once it’s messed up won’t really help. There’s been a bit of movement in this code since DP 2 so it may be that your issue is fixed already. If you see it again do you think you could take some screen recording of it and send to me?
> 
> I'd need to find some screen recording software. Preferably free, if you know of anything.

If you are on a Mac then you can use quicktime player.
> 
>>> We probably also need a way to turn off remote debugging when we're done with it, just in case.
>> Remote debugging might be a bit of a red herring here. It shouldn’t be doing anything unless there’s stuff coming in over the socket.
> 
> I could be falsely assigning cause and effect. Most of my session involved testing an app on a tethered phone, so debugging was the main thing that was different, and I had no trouble editing the script until after I started remote debugging it. But it could certainly be something else.

Well… I’m not sure at this stage but I _really_ want to fix it so if we can come up with a recipe it would be great. I’ll review the text entry code again to see if I can spot anything.

Cheers

Monte