repl needs work

[rdm]

I am unable to copy and paste from my session after entering these commands:

13!:0]1
i.3 3
example=:{{1 2+y}}
example i.3 3

The artifacts left in the session here, I think, are interesting, and shed light on some important issues which should be addressed.

[razetime]

when i run these commands from the editor, I get an endless stream of alert popups. Is that what you are also facing?

[rdm]

Yes.

[plj541]

I just noticed there was a page 2 on the issues list. I don't know enough about the internals of J to be sure, but the list of problems is the same as the initial release on Android. You cannot:
enter statements after a error once debugging is on 13!:0 [1
create multiline definitions in term
1!:1 [1 doesn't work
In Android, that was fixed with
wd 'setj asyncj 1'
The need for which may have been removed in subsequent releases.

[bilam]

I'm not sure but maybe setj asyncj 1 had become the default in j android. The disadvantage of asyncj is that the opengl demo can't run.

[plj541]

I remember there was a caution that something wouldn't run, but I had forgotten what it was. Are the pretty graphs j-playground has today opengl?

I was hoping the J dll could recognize the condition and provide an unusual prompt like
1 2 3{a.
Which the browser code could check for and discard, but know that the last request is
finished and its ok to send another.

I've also worked out a J solution using Javascript prompt for taking user requests during
debugging, but the system doesn't show my screen writes until the debugging has completed.

[plj541]

The contents of this file needs work and is little more than a proof-of-concept at this point.
https://plj541.github.io/Tools/Debug.txt
I've had its predecessor working on my machine for days, but until today debugging was blind on
J-playground
Please let me know what you think.

[plj541]

It's our looks like somebody else is where I was yesterday, only their code is installed. I don't know that JavaScript alert is good enough, but I found no other way to see what's happening.

[rdm]

In the current build setting a stop in emj.wasm on line 0x0122fa5 (a call_indirect() with a bunch of parameters which might be zeros) intercepts control immediately before the javascript popup (after logging the J error message to the javascript console). So presumably that's what we need to fix.