[lit-ssr] Initial commit of lit-ssr package #1347

kevinpschaaf · 2020-10-03T01:51:33Z

Port of SSR implementation from proof-off-concept work in lit-ssr repo, lit-html#hydration branch, and lit-element#hydration branch.

Current status:

Server-only render-lit-html tests passing
lit-html integration tests passing
LitElement integration tests passing
- renderLight directive implemented / tests passing
SSR unit/integration passing on CI

Note that the work under #1360 (additional fixes/features for end-to-end app proof-of concept) will be done in a follow-on PR.

Current size comparison: (as of b3c0d39)

Name                            Size      Minified  Gzipped  Brotli   
----------------------------------------------------------------------
lit-html.js  (lit-next)         6.41 KB   6.41 KB   2.81 KB  2.55 KB  
lit-html.js  (lit-next-ssr)     6.76 KB   6.76 KB   2.96 KB  2.67 KB (+150b gzip)

hydrate.js                      2.79 KB   2.79 KB   1.29 KB  1.15 KB  
private-ssr-support.js           168 B     167 B     159 B    127 B    

lit-element.js (lit-next)        694 B     693 B     394 B    330 B    
lit-element.js (lit-next-ssr)    810 B     809 B     431 B    372 B  (+37b gzip)   

hydrate-support.js               387 B     386 B     222 B    177 B

Notes for reviewers:

The first commit was a copy/paste of lit-ssr and hydrate.ts from the POC repo/branches, so https://github.com/Polymer/lit-html/compare/938459a..b3c0d39 will compare from that baseline forward to just before the formatting changes added at the end that make it much more difficult to diff
lit-html
- Added export const $private = { ... } export to lit-html which exports private top-level members needed by hydrate & render-lit-html as mangled names, which both keeps file size down and reinforces private nature of them
- Added private-ssr-support.js module which re-exports the fields in $private back to well-known symbol names for use in lit-ssr, so that package is agnostic to whether the lit packages are in dev or prod mode
- Added renderLight under lit-html /directives; we had this in LitElement, but it's not really LitElement-specific per se, and it fits better under lit-html directives than as a one-off under LitElement
- During hydration, we need to construct a NodePart before we've reached its end marker; for now I've added a _setEndNode method rather than make _endNode public; TBD
- In order to reuse the non-trivial AttributePart logic for running directives on all of the parts and concatenating everything together, I factored _resolveValue out of _setValue, so that we can call that during hydrate and on the server without committing the value to the DOM. However, as part of that we need to do the server thing with directives, aka run its render not its update. In order to reuse the AttributePart logic, I made Directive have a patchable resolve method that defaults to running update but which can be patched by SSR to run render instead. This is maybe one of the more questionable changes. TBD
lit-element
- Added hydrate-support.js module which installs the hydrate codepath into LitElement
- Adds _renderImpl to LitElement as a patching point for hydrate-support (similar to the old static render method, but internal and on the prototype)
template-shadowroot
- For now, I'm depending on the github repo and building it package from source as part of the lit-ssr build script since it's not released. We should discuss whether to publish a release yet or not.
lit-ssr
- I forgot to mention, I meant to stage these in separately, but I actually copy/pasted lit-ssr from a branch containing following PR's which were technically not yet merged into lit-ssr, so it's worth scrutinizing these changes also:
  - Updates for integration with lit-ssr-hacker-news app PolymerLabs/lit-ssr#64: dom-shim improvements + Koa middleware
  - Add support for CE attribute rendering. PolymerLabs/lit-ssr#66: Add support for CE attribute rendering

github-actions · 2020-10-05T19:04:41Z

📊 Tachometer Benchmark Results

Summary

⏱ lit-element-stub1

nop-update

lit-html-kitchen-sink: unsure 🔍 -4% - +3% (-1.45ms - +1.25ms)
^{this-change vs tip-of-tree}

render

lit-element-list: faster ✔ 1% - 2% (0.64ms - 2.31ms)
^{this-change vs tip-of-tree}
lit-html-kitchen-sink: unsure 🔍 -2% - +1% (-0.69ms - +0.38ms)
^{this-change vs tip-of-tree}
lit-html-repeat: unsure 🔍 -1% - +1% (-0.17ms - +0.17ms)
^{this-change vs tip-of-tree}
lit-html-template-heavy: slower ❌ 2% - 4% (1.25ms - 3.07ms)
^{this-change vs tip-of-tree}
updating-element-list: unsure 🔍 -2% - +1% (-1.03ms - +0.32ms)
^{this-change vs tip-of-tree}

update

lit-element-list: unsure 🔍 -0% - +1% (-3.91ms - +9.58ms)
^{this-change vs tip-of-tree}
lit-html-kitchen-sink: slower ❌ 1% - 7% (0.70ms - 7.62ms)
^{this-change vs tip-of-tree}
lit-html-repeat: unsure 🔍 -2% - +0% (-6.80ms - +1.17ms)
^{this-change vs tip-of-tree}
lit-html-template-heavy: unsure 🔍 -2% - +1% (-3.16ms - +1.04ms)
^{this-change vs tip-of-tree}
updating-element-list: unsure 🔍 -2% - +2% (-2.71ms - +2.17ms)
^{this-change vs tip-of-tree}

update-reflect

lit-element-list: unsure 🔍 -0% - +1% (-3.89ms - +11.30ms)
^{this-change vs tip-of-tree}
updating-element-list: unsure 🔍 -1% - +1% (-2.77ms - +1.51ms)
^{this-change vs tip-of-tree}

Results

lit-element-list

Browser: chrome-headless 86.0.4240.111
Sample size: 50
Built by: Benchmarks #449
Commit: f1b0ec1

render

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	106.42ms - 107.54ms	-	faster ✔ 1% - 2% 0.64ms - 2.31ms	faster ✔ 22% - 24% 30.62ms - 32.89ms
tip-of-tree tip-of-tree	107.84ms - 109.08ms	slower ❌ 1% - 2% 0.64ms - 2.31ms	-	faster ✔ 21% - 23% 29.11ms - 31.44ms
previous-release previous-release	137.75ms - 139.72ms	slower ❌ 29% - 31% 30.62ms - 32.89ms	slower ❌ 27% - 29% 29.11ms - 31.44ms	-

update

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	981.98ms - 992.13ms	-	unsure 🔍 -0% - +1% -3.91ms - +9.58ms	faster ✔ 8% - 9% 81.01ms - 94.16ms
tip-of-tree tip-of-tree	979.78ms - 988.66ms	unsure 🔍 -1% - +0% -9.58ms - +3.91ms	-	faster ✔ 8% - 9% 84.32ms - 96.52ms
previous-release previous-release	1070.45ms - 1078.83ms	slower ❌ 8% - 10% 81.01ms - 94.16ms	slower ❌ 9% - 10% 84.32ms - 96.52ms	-

update-reflect

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	1095.12ms - 1106.70ms	-	unsure 🔍 -0% - +1% -3.89ms - +11.30ms	faster ✔ 3% - 5% 36.51ms - 52.19ms
tip-of-tree tip-of-tree	1092.29ms - 1102.13ms	unsure 🔍 -1% - +0% -11.30ms - +3.89ms	-	faster ✔ 4% - 5% 40.83ms - 55.27ms
previous-release previous-release	1139.98ms - 1150.54ms	slower ❌ 3% - 5% 36.51ms - 52.19ms	slower ❌ 4% - 5% 40.83ms - 55.27ms	-

⏱ lit-element-stub1

Currently running in Benchmarks #281…

lit-html-kitchen-sink

Browser: chrome-headless 86.0.4240.111
Sample size: 30
Built by: Benchmarks #449
Commit: f1b0ec1

render

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	43.71ms - 44.24ms	-	unsure 🔍 -2% - +1% -0.69ms - +0.38ms	faster ✔ 20% - 24% 10.83ms - 13.81ms
tip-of-tree tip-of-tree	43.67ms - 44.59ms	unsure 🔍 -1% - +2% -0.38ms - +0.69ms	-	faster ✔ 19% - 24% 10.63ms - 13.71ms
previous-release previous-release	54.83ms - 57.76ms	slower ❌ 25% - 31% 10.83ms - 13.81ms	slower ❌ 24% - 31% 10.63ms - 13.71ms	-

update

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	112.03ms - 117.73ms	-	slower ❌ 1% - 7% 0.70ms - 7.62ms	unsure 🔍 -4% - +2% -4.75ms - +2.40ms
tip-of-tree tip-of-tree	108.76ms - 112.68ms	faster ✔ 1% - 7% 0.70ms - 7.62ms	-	faster ✔ 2% - 7% 2.43ms - 8.25ms
previous-release previous-release	113.90ms - 118.21ms	unsure 🔍 -2% - +4% -2.40ms - +4.75ms	slower ❌ 2% - 8% 2.43ms - 8.25ms	-

nop-update

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	38.48ms - 40.28ms	-	unsure 🔍 -4% - +3% -1.45ms - +1.25ms	faster ✔ 4% - 13% 1.50ms - 5.83ms
tip-of-tree tip-of-tree	38.47ms - 40.49ms	unsure 🔍 -3% - +4% -1.25ms - +1.45ms	-	faster ✔ 3% - 13% 1.35ms - 5.78ms
previous-release previous-release	41.07ms - 45.01ms	slower ❌ 4% - 15% 1.50ms - 5.83ms	slower ❌ 3% - 15% 1.35ms - 5.78ms	-

lit-html-repeat

Browser: chrome-headless 86.0.4240.111
Sample size: 30
Built by: Benchmarks #449
Commit: f1b0ec1

render

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	13.81ms - 14.02ms	-	unsure 🔍 -1% - +1% -0.17ms - +0.17ms	faster ✔ 19% - 21% 3.22ms - 3.57ms
tip-of-tree tip-of-tree	13.79ms - 14.05ms	unsure 🔍 -1% - +1% -0.17ms - +0.17ms	-	faster ✔ 19% - 21% 3.20ms - 3.59ms
previous-release previous-release	17.17ms - 17.45ms	slower ❌ 23% - 26% 3.22ms - 3.57ms	slower ❌ 23% - 26% 3.20ms - 3.59ms	-

update

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	424.36ms - 429.75ms	-	unsure 🔍 -2% - +0% -6.80ms - +1.17ms	faster ✔ 30% - 32% 186.42ms - 198.29ms
tip-of-tree tip-of-tree	426.94ms - 432.80ms	unsure 🔍 -0% - +2% -1.17ms - +6.80ms	-	faster ✔ 30% - 31% 183.50ms - 195.58ms
previous-release previous-release	614.13ms - 624.70ms	slower ❌ 44% - 47% 186.42ms - 198.29ms	slower ❌ 43% - 46% 183.50ms - 195.58ms	-

lit-html-template-heavy

Browser: chrome-headless 86.0.4240.111
Sample size: 50
Built by: Benchmarks #449
Commit: f1b0ec1

render

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	77.71ms - 79.03ms	-	slower ❌ 2% - 4% 1.25ms - 3.07ms	faster ✔ 15% - 18% 14.12ms - 16.71ms
tip-of-tree tip-of-tree	75.58ms - 76.84ms	faster ✔ 2% - 4% 1.25ms - 3.07ms	-	faster ✔ 18% - 20% 16.29ms - 18.86ms
previous-release previous-release	92.66ms - 94.90ms	slower ❌ 18% - 21% 14.12ms - 16.71ms	slower ❌ 21% - 25% 16.29ms - 18.86ms	-

update

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	167.27ms - 170.44ms	-	unsure 🔍 -2% - +1% -3.16ms - +1.04ms	faster ✔ 12% - 14% 23.40ms - 27.87ms
tip-of-tree tip-of-tree	168.53ms - 171.29ms	unsure 🔍 -1% - +2% -1.04ms - +3.16ms	-	faster ✔ 12% - 14% 22.49ms - 26.67ms
previous-release previous-release	192.91ms - 196.06ms	slower ❌ 14% - 17% 23.40ms - 27.87ms	slower ❌ 13% - 16% 22.49ms - 26.67ms	-

updating-element-list

Browser: chrome-headless 86.0.4240.111
Sample size: 50
Built by: Benchmarks #449
Commit: f1b0ec1

render

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	58.73ms - 59.49ms	-	unsure 🔍 -2% - +1% -1.03ms - +0.32ms	unsure 🔍 -1% - +1% -0.73ms - +0.38ms
tip-of-tree tip-of-tree	58.91ms - 60.02ms	unsure 🔍 -1% - +2% -0.32ms - +1.03ms	-	unsure 🔍 -1% - +1% -0.51ms - +0.86ms
previous-release previous-release	58.89ms - 59.69ms	unsure 🔍 -1% - +1% -0.38ms - +0.73ms	unsure 🔍 -1% - +1% -0.86ms - +0.51ms	-

update

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	129.14ms - 133.12ms	-	unsure 🔍 -2% - +2% -2.71ms - +2.17ms	unsure 🔍 -2% - +2% -2.55ms - +2.44ms
tip-of-tree tip-of-tree	129.99ms - 132.81ms	unsure 🔍 -2% - +2% -2.17ms - +2.71ms	-	unsure 🔍 -1% - +2% -1.85ms - +2.27ms
previous-release previous-release	129.68ms - 132.69ms	unsure 🔍 -2% - +2% -2.44ms - +2.55ms	unsure 🔍 -2% - +1% -2.27ms - +1.85ms	-

update-reflect

Version	Avg time	vs this-change	vs tip-of-tree tip-of-tree	vs previous-release previous-release
this-change	207.99ms - 211.22ms	-	unsure 🔍 -1% - +1% -2.77ms - +1.51ms	unsure 🔍 -0% - +2% -0.47ms - +3.55ms
tip-of-tree tip-of-tree	208.84ms - 211.64ms	unsure 🔍 -1% - +1% -1.51ms - +2.77ms	-	slower ❌ 0% - 2% 0.34ms - 4.01ms
previous-release previous-release	206.88ms - 209.26ms	unsure 🔍 -2% - +0% -3.55ms - +0.47ms	faster ✔ 0% - 2% 0.34ms - 4.01ms	-

_{tachometer-reporter-action v2 for Benchmarks}

Fix lit-element import paths For now, build template-shadowroot as part of lit-ssr Add lit-ssr tests to CI Fix merge error (Template:_options) Remove tsconfig.tsbuildinfo Uncomment element tests ffrom server-only suite Fix return value of createRenderRoot patch

Rename $litPrivate -> $private Fix demo hydration README updates

Some minor cleanup included

kevinpschaaf · 2020-10-16T09:16:11Z

packages/lit-ssr/src/lib/render-lit-html.ts

+          });
+          // There will be two comments per part (open+close) in the rendered
+          // output and on the client, so increment again for that
+          // nodeIndex++;


Need to confirm why this is no longer true, but likely due to the client-side change to only make one marker for node parts.

kevinpschaaf · 2020-10-16T09:17:47Z

packages/lit-html/src/lit-html.ts

  abstract render(...props: Array<unknown>): unknown;
  update(_part: Part, props: Array<unknown>): unknown {
    return this.render(...props);
  }
 }

+const getTemplateHtml = (strings: TemplateStringsArray, type: ResultType) => {


This is just broken out of the constructor, for use by render-lit-html.

kevinpschaaf · 2020-10-16T09:19:17Z

packages/lit-html/src/lit-html.ts

@@ -319,15 +322,156 @@ const walker = d.createTreeWalker(
 * change in future pre-releases.
 */
 export abstract class Directive {
+  resolve(part: Part, props: Array<unknown>): unknown {
+    return this.update(part, props);
+  }


In order to reuse the non-trivial AttributePart logic for running directives on all of the parts and concatenating everything together, I factored _resolveValue out of _setValue, so that we can call that during hydrate and on the server without committing the value to the DOM. However, as part of that we need to do the server thing with directives, aka run its render not its update. In order to reuse the AttributePart logic, I made Directive have a patchable resolve method that defaults to running update but which can be patched by SSR to run render instead. We can't just patch update because the user overrides this method. This is maybe one of the more questionable changes. TBD

Rename to _resolve and make private?

Done (it's @internal, not private)

Without having thought too hard about his yet, I'd be a little inclined to see if we can vary the callsite for directive.update() instead.

kevinpschaaf · 2020-10-16T09:20:56Z

packages/lit-html/src/lit-html.ts

+  /** @internal */
+  _setEndNode(node: ChildNode | null) {
+    this._endNode = node;
+  }


During hydration, we need to construct a NodePart before we've reached its end marker; for now I've added a _setEndNode method rather than make _endNode public; TBD

I think _startNode and _endNode should just be @internal.

Unless we do a rollup build for SSR (which we discussed in chat and decided to go the private-support approach for top-level exports), any mangled fields of those members still hit the issue, and need to be in the reserved list (not the cross-package mangle list). So they're both not hidden and not mangled. It's just a choice we need to decide on and then be consistent.

Sorry I take that back, _startNode/_endNode are only used in hydrate, which can be mangled, so I'll go for that.

packages/lit-html/src/directives/render-light.ts

kevinpschaaf · 2020-10-16T09:22:48Z

packages/lit-ssr/package.json

+  "description": "",
+  "main": "index.js",
+  "scripts": {
+    "build": "(cd node_modules/template-shadowroot && npm run build) && tsc",


For now, I'm depending on the github repo for template-shadowroot and building it package from source as part of the lit-ssr build script since it's not released. We should discuss whether to publish a release yet or not.

Unless something has changed recently with the spec'd feature that requires the package to be updated, I think we should release it.

Agreed. We can do a pre-release version for now.

packages/lit-ssr/package.json

Co-authored-by: Abdón Rodríguez Davila <a@abdonrd.com>

Return `noChage` from default render now that we don't clear the container before rendering, which was the reason for renderNotImplemented.

kevinpschaaf · 2020-10-22T08:00:05Z

@justinfagnani @sorvell I think I addressed most of the feedback in here, PTAL.

sorvell · 2020-10-22T21:58:22Z

packages/lit-element/src/hydrate-support.ts

+  new (...args: any[]): PatchableLitElement;
+  createRenderRoot(): Element | ShadowRoot;
+  _needsHydration: boolean;
+  _renderImpl(


packages/lit-html/src/directives/render-light.ts

packages/lit-html/src/hydrate.ts

sorvell · 2020-10-22T22:05:43Z

packages/lit-html/src/hydrate.ts

+  (container as any).$lit$ = rootPart;
+};
+
+const openNodePart = (


Explain that after calling hydrate lit-html will behave as if it just initially rendered that DOM and any subsequent renders will update efficiently, the same as if lit-html had rendered the DOM on the client.

packages/lit-html/src/hydrate.ts

packages/lit-ssr/src/lib/render-lit-html.ts

packages/lit-ssr/src/lib/util/parse5-utils.ts

packages/lit-ssr/src/test/integration/client/basic_test.ts

abdonrd · 2020-10-23T08:30:43Z

.github/workflows/tests.yml

+      - name: Lint
+        run: npm run lint
+


Suggested change

- name: Lint

run: npm run lint

Now lit-next branch already has a lint step! 🙂

kevinpschaaf · 2020-10-23T18:01:20Z

@sorvell Added comments & fixed attributePart naming. PTAL: 14837d4

justinfagnani · 2020-10-23T18:21:18Z

packages/lit-element/src/lit-element.ts

@@ -139,10 +130,14 @@ export class LitElement extends UpdatingElement {
   * the element to update.
   */
  protected render(): unknown {
-    return renderNotImplemented;
+    return noChange;


is this an intended change?

Ah right, that deserves some discussion. @sorvell had suggested just re-implementing LitElement's update in hydrate-support rather than adding a patchable _renderImpl, since LitElement's update is pretty small, and that would have required exporting the renderNotImplemented symbol. But in thinking about why we had that, it was to disambiguate whether the user implemented render so that we didn't clear the light DOM if the user had not implemented render. Since we don't clear the container now, I don't think we need to treat "not implemented" special anymore, we can just return noChange to not do anything in render. Seems better?

FWIW, this was the change where renderNotImplemented was introduced: lit/lit-element#917:

lit/lit-element#712 Introduced a breaking behavior change for situations where
render is unimplemented, and DOM is added before being connected to
the document.

justinfagnani · 2020-10-23T18:22:25Z

packages/lit-html/src/lit-html.ts

@@ -35,7 +35,7 @@ const markerMatch = '?' + marker;
 // syntax because it's slightly smaller, but parses as a comment node.
 const nodeMarker = `<${markerMatch}>`;

-const d = document;
+const d = window.document;


I don't think this should be necessary since we patch the global. Is this for the non-importModule use case?

Ah, we don't actually add the document instance to the shim. Lemme do that.

justinfagnani · 2020-10-23T18:22:49Z

packages/lit-html/src/lit-html.ts

-);
+const walker =
+  d &&
+  d.createTreeWalker(


d?.createWalker

Done (removed, added createTreeWalker stub to shim)

justinfagnani · 2020-10-23T18:24:53Z

packages/lit-html/src/lit-html.ts

@@ -319,15 +322,156 @@ const walker = d.createTreeWalker(
 * change in future pre-releases.
 */
 export abstract class Directive {
+  resolve(part: Part, props: Array<unknown>): unknown {
+    return this.update(part, props);
+  }


Without having thought too hard about his yet, I'd be a little inclined to see if we can vary the callsite for directive.update() instead.

justinfagnani · 2020-10-23T18:25:25Z

packages/lit-html/src/lit-html.ts

+  return {
+    // We don't technically need to close the SVG tag since the parser
+    // will handle it for us, but the SSR parser doesn't like that
+    html: html + strings[l] + (type === SVG_RESULT ? '</svg>' : ''),


can these names be minified?

Right now we don't have a great way to minify object/class fields that are used in SSR (just top-level exports via the $private export). Since this is a very private API, seems chill enough to return an array ([html, attrNames]), so I'll do that.

justinfagnani · 2020-10-23T22:18:45Z

packages/lit-html/src/lit-html.ts

@@ -751,7 +784,7 @@ export class NodePart {
    // TODO (justinfagnani): To support nested directives, we'd need to
    // resolve the directive result's values. We may want to offer another
    // way of composing directives.
-    this._setValue(this._directive.update(this, value.values));
+    this._setValue(this._directive._resolve(this, value.values));


How much benefit do you see out of reusing the part classes in their entirety? If we moved some of the methods to functions, could we use them to build SSR-specific implementations that call directive.render()?

So, in SSR we only construct AttributePart (the code to run the directive for NodePart is chill enough that I just copied it). For AttributePart, we basically want to reuse _resolveValue, which sequentially resolves each directive, deals with the nothing/noChange, and then concatenates the new string. Even if we made that into a function, we'd still need to thread a flag or function through to do something different on the server. Not sure that's worth it?

sorvell

LGTM

…Value override

…ue during SSR

kevinpschaaf · 2020-10-30T01:19:22Z

Ok, so I've confirmed that any kitchen-sink-update regression is now fully explained by adding directive support to EventPart; the change below results in the benchmark coming out unsure:

We'll look into optimizing directives in a separate issue: #1398.

sorvell · 2020-10-30T01:56:43Z

packages/lit-html/src/hydrate.ts

@@ -359,12 +357,11 @@ const createAttributeParts = (
      // directive value; we only then commit that value for event/property
      // parts since those were not serialized (passing `undefined` uses the
      // default commitValue; passing `noOpCommit` bypasses it)
-      const commitValue =
+      const noCommit = !(


Update the comment above this.

kevinpschaaf force-pushed the lit-next-ssr branch from 114aaac to e939aff Compare October 5, 2020 19:04

kevinpschaaf force-pushed the lit-next-ssr branch 2 times, most recently from 6453779 to 4c586dc Compare October 14, 2020 06:44

Initial SSR commit (code copied from #hydration & lit-ssr)

938459a

kevinpschaaf force-pushed the lit-next-ssr branch from 5aade07 to 5644c5a Compare October 15, 2020 23:50

kevinpschaaf added 6 commits October 15, 2020 17:00

Initial lit-html rendering/hydration (element tests disabled)

370344e

Add private-ssr-support to lit-html for re-exporting private symbols

e895d31

Cleanup & update README

51d6a58

Rename $litPrivate -> $private Fix demo hydration README updates

Add renderLight directive

3797d58

Disable incremental compilation for now

b3c0d39

kevinpschaaf force-pushed the lit-next-ssr branch from 5644c5a to b3c0d39 Compare October 16, 2020 00:22

kevinpschaaf added 2 commits October 16, 2020 00:24

Update ssr code to new lint rules

1036e3b

Some minor cleanup included

Format all and fix formatting errors

acd3914

kevinpschaaf commented Oct 16, 2020

View reviewed changes

packages/lit-html/src/directives/render-light.ts Show resolved Hide resolved

kevinpschaaf commented Oct 16, 2020

View reviewed changes

abdonrd reviewed Oct 16, 2020

View reviewed changes

packages/lit-ssr/package.json Outdated Show resolved Hide resolved

abdonrd reviewed Oct 16, 2020

View reviewed changes

packages/lit-ssr/package.json Outdated Show resolved Hide resolved

kevinpschaaf marked this pull request as ready for review October 16, 2020 09:27

kevinpschaaf requested a review from justinfagnani as a code owner October 16, 2020 09:27

kevinpschaaf requested a review from sorvell October 16, 2020 09:29

kevinpschaaf and others added 4 commits October 16, 2020 02:35

Remove format script from lit-ssr package.json

a4255d8

Co-authored-by: Abdón Rodríguez Davila <a@abdonrd.com>

Cleanup package.json

fa21d91

Add a strawman for files to publish

768d82a

Add test from PolymerLabs/lit-ssr#63

553f9cc

kevinpschaaf added 4 commits October 21, 2020 23:05

Switch to escape-html

7ceb334

Patch update instead of render

033abd5

Return `noChage` from default render now that we don't clear the container before rendering, which was the reason for renderNotImplemented.

Address more review feedback

2374fb1

Remove middleware, move to new PR

07d089b

sorvell requested changes Oct 22, 2020

View reviewed changes

abdonrd reviewed Oct 23, 2020

View reviewed changes

kevinpschaaf added 3 commits October 23, 2020 09:55

Add test for directive in EventPart

8cb9619

Merge branch 'lit-next' into lit-next-ssr

9d2df41

Address review feedback

14837d4

justinfagnani approved these changes Oct 23, 2020

View reviewed changes

kevinpschaaf added 3 commits October 23, 2020 16:03

Add licenses

ca7c666

Add a little more to dom-shim to remove client checks

668c152

Return an array from getTemplateHtml for terseness

8923601

sorvell approved these changes Oct 26, 2020

View reviewed changes

kevinpschaaf mentioned this pull request Oct 27, 2020

[lit-ssr] Investigate benchmarking impact of lit-ssr #1392

Closed

kevinpschaaf added 4 commits October 27, 2020 17:19

Merge branch 'lit-next' into lit-next-ssr

a5182c3

Revert _resolveValue factoring (had perf impact); use optional commit…

961e95c

…Value override

Downgrade Chrome for time being

3fc59a1

Try the other direction with Chrome

c7163e0

kevinpschaaf mentioned this pull request Oct 29, 2020

[lit-html] Explore perf optimization of directive resolution #1398

Open

kevinpschaaf added 2 commits October 29, 2020 17:51

Switch to using a boolean for overriding commit, and patch _commitVal…

ece2896

…ue during SSR

Merge branch 'lit-next' into lit-next-ssr

49a7f28

sorvell reviewed Oct 30, 2020

View reviewed changes

sorvell approved these changes Oct 30, 2020

View reviewed changes

Revise comment

458be70

kevinpschaaf merged commit dcd088f into lit-next Oct 30, 2020

kevinpschaaf deleted the lit-next-ssr branch October 30, 2020 16:30

[lit-ssr] Initial commit of lit-ssr package #1347

[lit-ssr] Initial commit of lit-ssr package #1347

Conversation

kevinpschaaf commented Oct 3, 2020 • edited

github-actions bot commented Oct 5, 2020 • edited

📊 Tachometer Benchmark Results

Summary

nop-update

render

update

update-reflect

Results

render

update

update-reflect

render

update

nop-update

render

update

render

update

render

update

update-reflect

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kevinpschaaf Oct 16, 2020 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kevinpschaaf commented Oct 22, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kevinpschaaf commented Oct 23, 2020 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sorvell left a comment

Choose a reason for hiding this comment

kevinpschaaf commented Oct 30, 2020

Choose a reason for hiding this comment

kevinpschaaf commented Oct 3, 2020 •

edited

github-actions bot commented Oct 5, 2020 •

edited

kevinpschaaf Oct 16, 2020 •

edited

kevinpschaaf commented Oct 23, 2020 •

edited