False error from KaSa: mod, tokens, `<-`

[hmedina]

%agent: timer(s{tik, tok})		%init: 1 timer()
%token: io		%init: 0 io		%obs: 'io_o' |io|
%token: oi		%init: 0 oi		%obs: 'oi_o' |oi|

timer(s{tik/tok}) | 1 io @ 10
timer(s{tok/tik}) | 1 oi @ 10
%mod: [T] > 5 do io <- 0 ;
%mod: [T] > 10 do $STOP;

The UI static analyzer reports 1/1[model.ka] Incorrect beginning of sentence , with a red squiggly under the io in the first mod.

Ignoring this message and hitting play produces the expected model behavior.

This syntax was referred by #607
The verbose version of the syntax is tested in https://github.com/Kappa-Dev/KappaTools/blob/master/tests/integration/compiler/tokens/tokens.ka

[hmedina]

%mod: [T] > 5 do $SNAPSHOT ; io <- 0 ;

With compound perturbations, weird things occur. The snapshot is not present; and KaSim CLI will refuse to run the model...

[feret]

Dear Hector,
The UI has never been adapted to KaSa error messages.
Most of them are badly reported without any relation with the alarm sent.
I will have to sort out and filter most of them out, while trying to understand how the UI is working.

[feret]

%mod: [T] > 5 do io <- 0 ;
is rejected by KaSim, not KaSa.

[feret]

This construction is not compliant with the klexer4/kparser4.
I have to update them accordingly.

[feret]

I extended the lexer and the parser in a branch.
Then, the issue is that it created a shift/reduce conflict in the grammar.
I am not sure I know how to change it without changing the syntax.
Adding a keyword %foo would solve the issue.

Would it be OK with you @hmedina ?
If someone knows how to deal with the parser to remove the shift/reduce conflict, please feel free to contribute.

[hmedina]

It would be ok with my usage (assuming it's just adding some keywords to make the "update token Y to value X" statement less cryptic...)

To elaborate, since the UI rejected that syntax, none of my big models use tokens & perturbations, because I did not know what a valid syntax would be. I only stumbled into working syntax when building a comprehensive KaTIE test, so changing the syntax will only cost me the time to re-write tests here and there.

The maintainers for both Kappa manuals will need to update their respective works once the new syntax is defined.

[feret]

I have solved the conflict by restricting a little bit the grammar.
Now
token <- alg
assignments are allowed if

a) the perturbation has a ending condition (repeat [ ])
as in %mod: [T] > 5 do io <- 0 repeat [true]

b) or if it is in a list of perturbations bounded by some parentheses.
as in %mod: [T] > 5 do (io <- 0)

Otherwise, it is not allowed.
For instance
%mod: [T] > 5 do io <- 0
is not OK.

[feret]

Since, the suggestion is strictly more expressive than what is in the master branch I merge.