@LuisMendo What was the first version of MATL to work with TIO?

@Suever TIO support started on January 14, 2016, with release 9.2.2

If you want to support multiple versions (I don't see it as essential, but it certainly would be a nice addition), you only need to keep .m and .mat files, including the compatibility folder. In particular, the .pdf and .txt files are relatively big and can be removed

Hm, maybe leave the .txt for completeness. It's the source file for all the function definitions

@LuisMendo Yea I've had the multiple releases baked in for a bit now and it's not that much overhead to make work. I have it auto-fetch the releases as people want to run something (if the release hasn't been downloaded previously). Also, any "save" link includes the version so that in theory the link will always link to code that runs regardless of future changes

@LuisMendo Oh no! My own bug report came back to bite us (changing char(0) to ' ')! You already overload disp (and inside of that call builtin('disp') so I can't detect disp calls

function disp(x)
% Avoids newline for '' input, or any output at all for [] input
%   Changes char(0) to space (Matlab displays char(0) as a space; Octave displays it as a space or
% as nothing, depending on platform)
if isempty(x)
    return
end
if ischar(x)
    x(x==0) = ' ';
end
builtin('disp', x);
end

@Suever :-) Maybe you can override (or is it overload, I'm never sure) disp by writing your own and placing it in a folder that's higher in the search path than compatibility?

oh ok so that gets pulled in dynamically?

@Suever Do you mean into the compiled code? No, it doesn't. The functions in the compatibility folder end up as subfunctions in the compiled file MATLc. I was suggesting to make MATLc see another disp function. But that won't work. You'd have to replace the function in compatibility for it to get pulled in

There must be a better way

I could get rid of the disp (sub)function and replace directly the code that implements D etc (move the [] and char(0) stuff there). That way there would not be a disp subfunction anymore in MATLc.m and you could override it.

Not a very nice approach, but it would work. I'll do that if we can't think of anything better

yea I'll play with it a little more.

It's not super critical except for the real-time functionality

for now I'm not worrying about that and just getting images in there

Maybe we should patch Octave's builtin so that builtin(disp) calls your disp :-D #NoWay #JustKidding

I actually thought of that....and then decided I'd rather blow my computer up

But as a joke I hope :-P

Hehe. Ok

yup

here's a bit of a teaser @LuisMendo

We really need to add the consonants to Y2 codegolf.stackexchange.com/a/83178/51939

@Suever \o/ Impressive!!

@Suever oh, what a pity

I will include them

A suggestion: maybe the version drop-down menu should just say "newest" by default, and have the others more hidden. Otherwise it might give the impression that having so many versions is a bit unfair advantage (extreme case is metagolfscript)

@LuisMendo haha yup that was exactly the plan was to say "latest" or something of the sort. Just hadn't gotten to it yet