feat(dia.CellView)!: separate alias and calc expression evaluation fro…

[kumilingus]

TBD

[alexandernst]

Is this related to #1421 ? 🤔

[kumilingus]

Currently, it allows you to switch off the support for camel-case. But if we move the alias logic to Vectorizer.attr() it would solve #1421.

[kumilingus]

The conversion of attribute names to kebab-case should be done efficiently. AFAIK there is no native API that uses camel-cased names as it is in the case of CSS properties e.g. `el.style.borderWidth = '5px'.

We can perhaps implement some caching like this:

const aliasMap = {
    xlinkHref: 'xlink:href',
    xlinkShow: 'xlink:show',
    xlinkRole: 'xlink:role',
    xlinkType: 'xlink:type',
    xlinkArcrole: 'xlink:arcrole',
    xlinkTitle: 'xlink:title',
    xlinkActuate: 'xlink:actuate',
    xmlSpace: 'xml:space',
    xmlLang: 'xml:lang',
    xmlBase: 'xml:base',
    preserveAspectRatio: 'preserveAspectRatio',
    requiredExtensions: 'requiredExtensions',
    requiredFeatures: 'requiredFeatures',
    systemLanguage: 'systemLanguage',
    externalResourcesRequired: 'externalResourcesRequired',
};

const aliases = new Proxy(aliasMap, {
    get(map, property) {
        if (property in map) {
            return map[property];
        } else {
            const alias = map[property] = util.toKebabCase(property);
            return alias;
        }
    }
});

Later, when using Vectorizer:

// get an alias
console.log(V.aliases.strokeWidh);

// setting attributes
V(el).attr('strokeWidth', 5);
V(el).attr({ strokeDashoffset': 100 });

// define custom / missing alias
V.aliases.myAlias = 'my:alias';

[kumilingus]

At this point, I'm not sure if it's worth optimizing. See https://jsperf.app/disifi.

[alexandernst]

🤔 it’s 11% slower on iPhone 12 mini, but indeed I don’t think it will really matter that much on desktop. Maybe on batch operations where multiple attrs of multiple cells will be modified?

[kumilingus]

Using caching was slower in all browsers on Desktop for me. Yep, more testing is required.

[kumilingus]

Obsolete now. See #2459.