Keypresses getting propagated from a launched program to Cesium when exiting

[andresesevilla]

Will capture last key press when closing programs

Ex:Open calcuzap,press clear, cesium will close.

[Eeems]

Can confirm that this happens in calcuzap. It must be something to do with how calcuzap handles the keypresses though, other programs don't cause cesium to close when pressing clear to exit.

[adriweb]

Might there be a way to clear the "key buffer" as soon as Cesium regains control, though?

[Eeems]

Looks like getkey is part of the main loop https://github.com/MateoConLechuga/Cesium/blob/master/src/routines/main.asm#L57
Waiting for all keys to be released before going back to the main loop might be a good idea. From what I can tell, it already is here https://github.com/MateoConLechuga/Cesium/blob/5bfb9bdd93aaf923492b4cef4d3cb371849acb6d/src/routines/common.asm#L38
It might be good to poke Patrick about how calcuzap exits to see if he's doing something strange.

[adriweb]

I have a fix: di before call DeletePgrmFromUserMem and ei before jp RELOADED_FROM_PRGM.
Confirmed to work with Calculzap.
Looks good?

[mateoconlechuga]

Um. If you enable interrupts, that means TIOS interrupts will be executing.
If it works, sure.

On Tue, Oct 13, 2015 at 4:52 PM, Adrien Bertrand notifications@github.com
wrote:

I have a fix: di before call DeletePgrmFromUserMem and ei before jp
RELOADED_FROM_PRGM.
Confirmed to work with Calculzap.
Looks good?

—
Reply to this email directly or view it on GitHub
#1 (comment)
.

[adriweb]

Hmm.
Well, I have pinged PatrickD on IRC, so we'll see what he thinks about his keypress handling related to that issue.

[Eeems]

You might want to just di. For all we know calcuzap enables interrupts Cesium wasn't handling them being on again properly.

EDIT: We may even want to di at ReturnHere https://github.com/MateoConLechuga/Cesium/blob/5bfb9bdd93aaf923492b4cef4d3cb371849acb6d/src/routines/reloader.asm#L44

[adriweb]

Yeah that looks safe enough.
I'll try that and commit soon.

[adriweb]

Actually, not doing ei doesn't fix the behaviour. Only di \ ... \ ei fixes it. I guess we'll have to go with that. So far, I haven't found any unexpected things with interrupts enabled at this point.
Same with a di at the ReturnHere location.

[adriweb]

Fixed in 1.1.2 (commit: 305c7ff)
Go redownload the binaries (8xp files) and give it a try.

Closing the issue for now.

[Eeems]

I've just tested and can confirm that this is fixed.