montagejs / frb Goto Github PK

groupMap and toMap don't seem to be working:

groupMap fribble example
toMap fribble example

If you change the property of one of the list items on which list was sorted this leads to unexpected behavior (some elements disappear and some are duplicated).

Unfortunatly mfiddle dosen't work properly so please see this gist for more details https://gist.github.com/anonymous/d03703f87917f8d6490c

I often have problems writing, understanding and debugging FRB expressions. Is there a public method that would look something like:

evaluateExpression(scope, expression)

For example:

evaluateExpression({
    folks: [
        {id: 4, name: "Bob"},
        {id: 2, name: "Alice"},
        {id: 3, name: "Bob"},
        {id: 1, name: "Alice"},
        {id: 1, name: "Alice"} // redundant
    ]
}, "folks.map{[id, this]}.toMap()") // {4: "Bob", 2: "Alice", 3: "Bob", 1: "Alice"}

It would help me enormously.

"If any operand is null or undefine, a binding will not emit an update. Thus, if a binding’s source becomes invalid, it does not corrupt its target but waits until a valid replacement becomes available."

This poses a problem when I expect the null or undefined to get passed along, to deal with the "corruption" at the endpoint.

e.g. In Montage I'd like some default handling for null or undefined src on the Image component but when a binding that had been a valid url becomes null the image never knows.

I have worked around this on a binding by binding basis by capping a.Potentially.Null.Path || nullPlaceholderValue

But this gets to be a little painful having to solve it on each binding.

I'd prefer to flip the convention to shield less-resilient objects from undefined/null while expecting objects to be tolerant by default.

@segphault observed a problem where a groupMap{} block worked, but a group{} block produced an empty array that would not update when the source array updated. This issue needs isolation and verification.

This is a useful undocumented feature:
https://github.com/montagejs/frb/blob/master/compile-evaluator.js#L233

Also needs to be renamed to evaluate, which may be breaking.

Thanks to @romancortes for discovering this.

Is there a specific reason this isn't already done? Adding two listeners just to get before and after seems avoidable. Looking through the code, I think I could get this to work, but wanted to check before I started the work.

I noticed this when exploring the addRangeAtPathChangeListener issue.

Here's the test case: https://gist.github.com/aadsm/5543418

It would be nice if I could set the produced regexp to be case-insensitive.

name.startsWith("mike")

to match: mike Mike mIKE etc.

Such that classList.(has("x"); has("y")) <-> z is equivalent to classList.has("x") <-> z; classList.has("y") <-> z, except that the bindings would be both directly asserted when classList changes, instead of independently.

I wanted to see what browsers this code supported so I tried to browserify the entry point.

➜  frb git:(master) ✗ browserify bindings.js -o bundle.js
Error: Parsing file /Users/paulgrenier/GitHub/frb/language.js: Line 31: Unexpected string

I was able to get through this by turning "if" into a constant. However, getting the tests to run is significantly harder--there are several references to other reserved words like 'with' and 'delete'.

Is there an expectation that these tests run on the client? If so, which browsers are supported?

I can't find any references to toMap on http://documentup.com/montagejs/frb/ or http://documentup.com/montagejs/collections/

In 0.2.12. Guess left over from local modifications

Evidence: montagejs/montagejs.org@02c4e9a#diff-33

The libraries are these:

<script src="https://cdn.jsdelivr.net/npm/nerdamer@latest/nerdamer.core.js"></script>
<script src="https://cdn.jsdelivr.net/npm/nerdamer@latest/Algebra.js"></script>
<script src="https://cdn.jsdelivr.net/npm/nerdamer@latest/Calculus.js"></script>
<script src="https://cdn.jsdelivr.net/npm/nerdamer@latest/Solve.js"></script>
<script src="https://cdn.jsdelivr.net/npm/[email protected]/index.min.js"></script>

The equation system thats works like a perceptron with that libraries is this:

[
    //perceptron

    'r1=b+(m1*x11)+(m2*x12)',
    'r2=b+(m1*x21)+(m2*x22)',
    'r3=b+(m1*x31)+(m2*x32)',

    //entrenamiento

    'r1=0',
    'x11=0',
    'x12=0',

    'r2=0',
    'x21=1',
    'x22=0',

    'r3=0',
    'x31=0',
    'x32=1',

    //soluciones

    'r4=b+(m1*x41)+(m2*x42)',

    'x41=0',
    'x42=0',

    'r5=b+(m1*x51)+(m2*x52)',

    'x51=1',
    'x52=0',

    'r6=b+(m1*x61)+(m2*x62)',

    'x61=0',
    'x62=1',

    'r7=b+(m1*x71)+(m2*x72)',

    'x71=1',
    'x72=1',
]

And the document doesnt print the solutions:

Any help? Thanks anyway!

In a component definition:

this.addPathChangeListener(
  "_element.parentElement.offsetWidth",
  this,
  "handleResize"
);

Will only trigger this.handleResize() once. offsetWidth is definitely changing by checking this way and resizing viewport:

setInterval((function () {
  console.log(this._element.parentElement.offsetWidth);
}).bind(this), 1000);

Is this possible?

The last time the grammar was compiled with pegjs was probably in 2013/2014. I don't believe our grammar changed since, but pegjs has. We should do a git bisect between the last version of pegjs that parses it correctly and now to identify what commit of pegjs introduce the issue so we can reach out to pegjs' author for some insight.