BaseObservable — Extracting lifecycle and attribute change observability from Y.Base

[ericf]

This extracts Y.BaseObservable, an extension for Y.BaseCore, from Y.Base.

By extracting the lifecycle and attribute change observability features, BaseCore subclasses have the potential to mix these features back in at runtime. These changes (and probably more) are needed to create the Y.Model.Base class and Y.Model.Observable extension.

I ran the base unit test and they all pass; additionally, I ran app's and a few other higher-level components to make sure I didn't introduce any gaps or regressions.

Running Locally

I explicitly did not commit the build files or the Loader metadata update, when trying out these changes locally, do the following:

$ cd src/attribute/ && shifter
$ cd ../base/ && shifter
$ cd ../yui/ && shifter

Todos

There are several things that still have to be done before this is merged in. Many of these came up in a code review of this Pull Request.

• Determine if `Y.BaseObservable` can be mixed into a `Y.BaseCore` subclass at runtime.

• Try implementing `Y.Model.Base` and `Y.Model.Observable` on this structure.

• Move Attribute's `_protecteAttrs()` prototype method to a static utility method on the `Y.Attribute` constructor function.

• Have `Y.BaseObservable` not mix `Y.Attribute` as a whole, instead:

  -

  `Y.Base` will itself mix-in `Y.AttributeExtras` before `Y.BaseObservable`.

  -

  Investigate whether `Y.Model.Base` should itself mix-in `Y.AttributeExtras`.

• Remove `Base._buildCfg = ` from `base-build`:

  -

  Create a separate object used for Base's config that's a copy of BaseCore's.

• Move the `_ATTR_CFG_HASH` computation into `base-build`.

• Naming: I kind of like "observable" better, e.g. `Y.BaseCore` and `Y.BaseObservable`. This should trickle down to `Y.AttributeObservable` too.

• Figure out correct yuidoc tags to use to make the API docs right (possibly a bug in yuidocjs).

[ericf]

Having this method on Y.Base may affect the ability to mix in Y.BaseEvents into a Y.BaseCore subclass at runtime.

[ericf]

@sdesai This is the initial work I've done to extract Y.BaseEvents from Y.Base. There is still more I want to do to make sure this services the needs of Y.Model.Base and Y.Model.Events. Let me know if anything stands out to you…

[lsmith]

I wonder if this is similar to the DT class extensions, in that I included @for DataTable.Base in the docs so the methods would show up where they were most likely to be searched for. The module description for the class extensions include a note that the API docs are in DataTable, since otherwise they appear empty. (e.g. http://yuilibrary.com/yui/docs/api/classes/DataTable.Mutable.html)

[ericf]

@lsmith Yup, this is the same thing I'm seeing. I guess we can sort that out when @davglass get's back and determine if there's something we want yuidoc to do differently or not.

[ericf]

I'm pretty sure _ATTR_CFG and _ATTR_CFG_HASH should be aggregated by Y.Base.create() and Y.Base.mix(). Looking at Y.Base._buildCfg, I see the following:

Base._buildCfg = {
    custom: {
        ATTRS: function (prop, r, s) { /* ... */ },
        _NON_ATTRS_CFG: arrayAggregator
    }
};

It seems natural to add _ATTR_CFG and possibly _ATTR_CFG_HASH to this set. This way, you could mix Y.Base.Events (which wants to add things to _ATTR_CFG) to a Y.Base.Core subclass at runtime via Y.Base.mix().

Since _ATTR_CFG_HASH depends on _ATTR_CFG being fully aggregated, it would be brittle to rely on object property order for its function to be called after _ATTR_CFG's :-/

[ericf]

The implementation I ended up with is in d92409c

[ericf]

@sdesai I'd like to get this in 3.x, but would like you to review it first. This is a precursor to #2532429, and @caridy might want to use the new capabilities this provides.

[caridy]

@sdesai the idea is to try to use this feature on the new mojito prototype we are building during the current cycle :)

[sdesai]

Will take a look this week.

[sdesai]

Sorry about the delay looking into this. I'm just going to put it in the next sprint, so it gets done.

[sdesai]

http://yuilibrary.com/projects/yui3/ticket/2532789

[ericf]

Updated the Todos list in the Pull Request description based on feedback from @sdesai when we had a code review.

[ericf]

Note: Travis is lying, this does actually pass all tests, I just decided against checking in the build files for the sake of the diff.

[davglass]

I'm not sure why my scripts aren't picking it up, I'll ping the Travis
guys to make sure the Env is getting set.

[ericf]

It has been determined that the API documentation anomalies are issues with YUIDoc that will be reported and fixed there.

[ericf]

I am going to merge this into the 3.x branch today. /cc @sdesai

[ericf]

This has been merged in 3.x.