Container: Add load / render / redraw lifecycle events

[50rayn]

Wanted a reliable way to know when a chart actually finished drawing — for logging first-paint times and reacting to renders downstream. The only existing hook was the onRenderComplete function-prop callback, which feels foreign in Vue/Svelte/Angular where events are the idiom. The usual workaround is await nextTick() + a couple of requestAnimationFrames, but that's a guess, not a real signal from unovis.

So I wired up load (first render), render (every render), redraw (every render except first) on VisXYContainer and VisSingleContainer — Highcharts-style. Vue gets @load, Svelte on:load, Angular (load). React/Solid keep onRenderComplete (function-prop callbacks are idiomatic there).

While at it, refactored the underlying XYContainer.onRenderComplete / SingleContainer.onRenderComplete from 7 positional args to a single payload object — that way XYContainerRenderPayload / SingleContainerRenderPayload live once in @unovis/ts and everyone imports them. This is a small breaking change for anyone passing onRenderComplete directly to the TS class or to React/Solid wrappers.

Verified pnpm check for Vue/Svelte, ng build for Angular, and the TS package rebuilds clean.

[rokotyan]

I like the idea! But there're a couple things I'm concerned about.

[rokotyan]

Switching to a payload object will be a breaking change, so we won't be able to do it till we release version 2.0. We'll need to stick to arguments

[50rayn]

Reverted the TS-core onRenderComplete to positional args; wrappers build XYContainerRenderPayload / SingleContainerRenderPayload internally before emitting. No breaking change for existing TS / React / Solid users.

[rokotyan]

Do you think we really need three of these events instead of just one? I think the users can track whether it was the first render themselves if they need to. Could you give a couple of use cases to justify that?

[50rayn]

I was inspired by Highcharts (chart.events.load / render / redraw). Since I'm currently migrating from it, I have a couple of moments where I use these. Use cases I had in mind:

• load — the big one. Fires once, perfect for prefetching chart settings, hydrating a "chart settings" panel, logging first-paint time, kicking off a fade-in, and hiding a skeleton. A userland boolean works, but every team ends up rewriting it and getting it subtly wrong (re-mounts, keyed components, hot reload, props that retrigger updates without a true "first paint").
• redraw — react only to data/prop updates without the noisy first paint. e.g. block a "chart settings" toggle button while a redraw is in flight, refit an overlay, restore scroll, re-sync an external legend.
• render — superset, blocks UI on every paint (first + redraw). Useful when you want a generic "chart is busy" gate around the whole lifecycle without branching on first vs subsequent.

Cost on our side is one boolean per container, so emitting all three is essentially free. Happy to drop redraw if it really feels redundant — but I'd push hard to keep load since it's the one consumers actually trip over. WDYT?