MATL CHATL

« first day (2309 days earlier) ← previous day next day → last day (894 days later) »

flawr

11:43

just got an edit from @Suever, apparently now it is available under matl.io !

3 hours later…

Luis Mendo

14:55

@flawr Yes, that's going to be the default domain now. The others will redirect to it

2 hours later…

Suever

17:00

Howdy @LuisMendo!

Looking into the reported issue now to see what's going on!

Also sorry everyone for the constant changing of domain names lol. matl.io is the last change, promise

Also sorry for all the post edit spam

Luis Mendo

17:37

@Suever Hey! Thank you for all the work, and for taking care of MATL Online

@Suever Let me know if I can help with that. Maybe I can look at the MATLc.m to see if I can find the issue

Suever

I was able to figure out what the issue is with Octave (I AM running a newer version than officially supported by MATL) and it seems to be a bug in octave (surprise!)

```octave:1> output = cell(1, 0);
octave:3> [output{:}] = colormap([1,0,0])
output =
{
[1,1] =

error: octave_base_value::print (): wrong type argument '<unknown type>'
octave:4> output = cell(1,1)
output =
{
[1,1] = [](0x0)
}

octave:5> [output{:}] = colormap([1,0,0])
output =
{
[1,1] =

1 0 0

}
```

It's getting confused by the {:} on an empty cell array in that it seems to set the size but doesn't actually fill it with anything which prevents it from even displaying properly

Luis Mendo

Huh! Octave bug, for variety :-)

If that only happens with colormap I can fix that easily in the compiler

Suever

@LuisMendo I'll check and see if it's fixed in a newer version real fast now that I can reproduce

Luis Mendo

But a newer Octave version wil likely cause more issues. How hard would it be for you to use Octave 4.2.2?

Suever

I wish we had an easy way to vett newer versions of octave

It isn't too terrible although I would have to switch to building octave from scratch because I need a new operating system for security with an old octave

Luis Mendo

18:03

If you can try output = cell(1, 0); [output{:}] = fun(...); with some typical fun other than colormap and it works, I can treat that as another Octave bug and patch it intercepting the colormap call. It would be specially easy if you can confirm that builtin('colormap', eye(3)) and a = builtin('colormap', eye(3)) work (calling the original colormap via builtin) in the newer Octave version you are using

Suever

Looks like it actually happens with all functions, not just colormap. builtin fixes it but widespread use of that could cause some side effects due to how the online interpreter works (by wrapping calls to specific functions)

I'm trying to see if I can get octave 4.2.2 to install from source which would be the safest option for now I think

Luis Mendo

18:21

Yes, the builtin approach is appropriate for specific functions, not for something so general. So it's a big bug in Octave, isn't it? Weird that it didn't get caught

What Octave version is it? Perhaps it has already been reported

Suever

@LuisMendo It's in 6.2, that's why I was looking to see if it was still an issue in newer versions because surely it was fixed by now

I think I have another solution where I can actually use two different versions. For the web frontend where all it does related to octave is generate the explanation I can use the newer octave. For the backend which actually runs the code I can use the older version that works without worrying about security since it isn't publicly accessible

Luis Mendo

19:05

@Suever There is 6.4.0 now but it still has the issue (I tried in octave-online.net)

I hope it's not too difficult to use two different versions

1 hour later…

flawr

20:22

@Suever And I do like it:)

1 hour later…

Luis Mendo

21:31

@Suever I have created a pull request (I think... I'm not good with GitHub) to replace the code for the "Game of Life" example by a golfier version. It's just to replace the current code in index.html by ?code=2%24rEk%60tZcXxDt3Y6QZ%2B5%3A7mT&inputs=20%0A50

2 hours later…

Luis Mendo

23:37

@flawr It's golfier, very appropriate :-)

« first day (2309 days earlier) ← previous day next day → last day (894 days later) »

all rooms