[Glimmer2] Implement input #13615

asakusuma · 2016-06-07T04:08:03Z

I don't think the late bound stuff can help with implementing input since lateBound gets called after CurlyComponentManager.create.

cc @krisselden @chadhietala @GavinJoyce

chadhietala · 2016-06-07T04:35:23Z

packages/ember-glimmer/lib/environment.js

@@ -146,6 +157,11 @@ export default class Environment extends GlimmerEnvironment {
          assert(`A helper named '${path[0]}' could not be found`, false);
        }
      } else {
+        let generateSyntax = builtInDynamicComponents[key];


Seems like this should be it's own branch?

Sure. It would just mean always doing the lookup up-front which seems fine.

asakusuma · 2016-06-07T04:54:03Z

For whatever reason, one of the input tests is failing in Safari and Microsoft browsers. I'm looking into it.

asakusuma · 2016-06-07T05:18:26Z

I think saucelabs is having problems

Error: 07 Jun 05:09:02 - All jobs using tunnel have finished.
07 Jun 05:09:02 - Sauce Connect could not establish a connection.
07 Jun 05:09:02 - Please check your firewall and proxy settings.

asakusuma · 2016-06-08T17:55:26Z

I've changed the approach to re-write the AST. Any ideas why travis doesn't seem to be running?

rwjblue · 2016-06-08T18:56:10Z

wat

rwjblue · 2016-06-08T19:47:15Z

packages/ember-glimmer-template-compiler/lib/plugins/transform-input-type-syntax.js

+    }
+  }
+  if (pair && pair.value.type !== 'StringLiteral') {
+    node.params.unshift(builders.sexpr('-input-type', [builders.path(pair.value.original)], null));


Please add loc information to these (builders.sexpr and builders.path).

What happens if -input-type is already specified?

@rwjblue how do I add loc information here? Neither buildSexpr nor buildPath accept a loc arguments like some of the other builders. Does that need to be added?

I don't think end-users can add a -input-type helper since it's registered as an internal helper. But if one already existed, I believe DynamicComponentSyntax only looks at the first argument, so if -input-type was already specified, it would be ignored.

Neither buildSexpr nor buildPath accept a loc arguments like some of the other builders. Does that need to be added?

Yes. Each AST node should be able to be built with a loc.

I don't think end-users can add a -input-type helper since it's registered as an internal helper.

There is nothing stopping a user from having already defined a -input-type helper in their application, but I'm not sure we should care about that too much...

I believe DynamicComponentSyntax only looks at the first argument, so if -input-type was already specified, it would be ignored.

If we expect this to go into exactly node.params[0] we should do that, and error if there was already something there.

For example today this "works" (doesn't blow up):

{{input "blah blah blah" value=someThing}}

If we are going to add (internally) significance to the first ordered argument to {{input, then we should have a helpful assertion if the user puts something there...

@rwjblue I meant that the end-user cannot use -input-helper unless they themselves define -input-helper. I confirmed this locally. My understanding is that a leading dash is reserved for internal helpers.

EDIT: In other words, the -input-helper I am introducing is not available for end users. They would have to make their own, which I'm under the impression is not supposed to happen given the convention.

I can certainly add a check to blow up if there's any pre-existing positional params. In other words, we are now saying that you cannot use positional parameters with {{input}}.

rwjblue · 2016-06-08T19:54:36Z

Looking good!

asakusuma · 2016-06-08T20:53:16Z

Working on figuring out why tests fail in Safari and MS browsers. Safari is failing for a different reason than MS.

asakusuma · 2016-06-08T21:39:58Z

Modifying input.selectionStart, which is utilized in the cursor tests, causes an event in Safari (does not happen in Chrome). This event kicks off a chain of events that results in an error because the view does not have sendAction implemented. Not sure if this is the right approach, but I solved this by turning off the event dispatcher for the input tests, since we are not interested in the event dispatcher for these tests.

chadhietala · 2016-06-10T05:21:05Z

@asakusuma Yea we need to land #13532.

homu · 2016-06-11T05:19:44Z

☔ The latest upstream changes (presumably #13492) made this pull request unmergeable. Please resolve the merge conflicts.

krisselden · 2016-06-14T19:37:56Z

@asakusuma Should you skip the failing test for now and we can recheck after sendAction?

asakusuma · 2016-06-14T20:17:22Z

I'm fine doing that, though I'm not sure sendAction is what's causing the Microsoft browser issues.

All the tests pass in Chrome, Safari, and Firefox. In Microsoft browsers, the null values test fails. What's failing is that when you set the value of the input to null, the value does not become an empty string in MS browsers for whatever reason.

homu · 2016-06-16T21:27:31Z

☔ The latest upstream changes (presumably #13684) made this pull request unmergeable. Please resolve the merge conflicts.

asakusuma · 2016-06-17T19:13:28Z

Discovered the source of the problem in MS Edge. removeAttribute('value'); doesn't seem to work in Edge. When the value needs to be null, glimmer uses removeAttribute('value'); to update the input element.

Here's a simple repro: https://gist.github.com/asakusuma/aad85f430f465903e56feee2b53a414c

cc @jdalton

rwjblue · 2016-06-17T22:17:38Z

@asakusuma - Have you reported an issue against MS Edge?

asakusuma · 2016-06-17T22:30:26Z

@jdalton in MS Edge, value is still "I have the best inputs" even after removeAttribute is run. On chrome, value attribute is removed and the input is empty.

I have not yet filed a ticket.

homu · 2016-06-18T06:00:48Z

☔ The latest upstream changes (presumably #13649) made this pull request unmergeable. Please resolve the merge conflicts.

jdalton · 2016-06-20T23:23:18Z

@asakusuma I filed a ticket internally.

rwjblue · 2016-06-20T23:30:06Z

@jdalton - Thank you!

jdalton · 2016-06-20T23:45:31Z

I've gotten removeAttribute to work for other attributes on other elements in Edge.
Do you all know of any other problem attributes?

asakusuma · 2016-06-20T23:59:39Z

@jdalton thanks! I'm not aware of any other problem attributes.

jdalton · 2016-06-21T00:14:44Z

Just a heads up, while making a feature detection for this I noticed that in Chrome:

var input = document.createElement('input');
var buggyRemoveAttrValue = !!(input.value=1,input.removeAttribute('value'),input.value);
buggyRemoveAttrValue;
// => true in Chrome and Edge

but

var div = document.createElement('div');
div.innerHTML = '<input value=1>';
var input = div.firstChild;

var buggyRemoveAttrValue = !!(input.removeAttribute('value'),input.getAttribute('value'));
buggyRemoveAttrValue;
// => false in Chrome but true in Edge

It's the old source attribute vs. element property thing.

homu · 2016-06-21T02:25:40Z

☔ The latest upstream changes (presumably #13725) made this pull request unmergeable. Please resolve the merge conflicts.

asakusuma · 2016-06-21T20:16:41Z

Did some bisecting, and it would appear that glimmerjs/glimmer-vm@951cbe5 introduced an incompatibility with the ember build.

Could not require 'ember-cli-build.js': Cannot find module 'glimmer-engine/ember-cli-build'
Error: Could not require 'ember-cli-build.js': Cannot find module 'glimmer-engine/ember-cli-build'
    at Function.Module._resolveFilename (module.js:339:15)
    at Function.Module._load (module.js:290:25)
    at Module.require (module.js:367:17)
    at require (internal/module.js:20:19)
    at Object.<anonymous> (/home/travis/build/emberjs/ember.js/ember-cli-build.js:45:21)
    at Module._compile (module.js:413:34)
    at Object.Module._extensions..js (module.js:422:10)
    at Module.load (module.js:357:32)
    at Function.Module._load (module.js:314:12)
    at Module.require (module.js:367:17)
    at require (internal/module.js:20:19)
    at module.exports (/home/travis/build/emberjs/ember.js/node_modules/ember-cli/lib/utilities/find-build-file.js:19:17)
    at CoreObject.module.exports.Task.extend.setupBroccoliBuilder (/home/travis/build/emberjs/ember.js/node_modules/ember-cli/lib/models/builder.js:48:21)
    at CoreObject.module.exports.Task.extend.init (/home/travis/build/emberjs/ember.js/node_modules/ember-cli/lib/models/builder.js:89:10)
    at CoreObject.superWrapper [as init] (/home/travis/build/emberjs/ember.js/node_modules/core-object/lib/assign-properties.js:32:18)
    at CoreObject.Class (/home/travis/build/emberjs/ember.js/node_modules/core-object/core-object.js:32:33)

cc @tomdale @rwjblue

tomdale · 2016-06-21T20:30:07Z

@asakusuma I think this is related to glimmerjs/glimmer-vm#190, we need to ship packages/ and ember-cli-build.js with the npm package for use by Ember.

asakusuma · 2016-06-21T21:10:39Z

After including the glimmer fix, Travis is now failing with a new problem

The Broccoli Plugin: [Funnel] failed with:
Error: ENOENT: no such file or directory, lstat '/home/travis/build/emberjs/ember.js/node_modules/glimmer-engine/node_modules/simple-html-tokenizer/lib'

asakusuma · 2016-06-22T02:51:30Z

@rwjblue do we need to increase the timeout?

Blueprints is also failing but I'm not seeing a useful error.

No output has been received in the last 10m0s, this potentially indicates a stalled build or something wrong with the build itself.

asakusuma · 2016-06-22T03:50:29Z

npm is still borking on the blueprints job. it keeps doing this which leads me to believe there is something wrong, but there's no useful error, just npm ERR! code 1

rwjblue · 2016-06-22T11:47:38Z

Cleared cache and restarted builds.

rwjblue · 2016-06-22T13:27:01Z

Thank you @asakusuma for sticking with this so long!

refs emberjs#13644

refs #13644

refs emberjs#13644

chadhietala reviewed Jun 7, 2016
View reviewed changes

asakusuma force-pushed the input-dynamic branch from b9d1926 to 78babd1 Compare June 7, 2016 05:01

asakusuma force-pushed the input-dynamic branch 3 times, most recently from ec51222 to dc776b4 Compare June 8, 2016 02:34

asakusuma force-pushed the input-dynamic branch from dc776b4 to 2978427 Compare June 8, 2016 19:10

rwjblue reviewed Jun 8, 2016
View reviewed changes

asakusuma force-pushed the input-dynamic branch from 2978427 to 4cb03d1 Compare June 8, 2016 21:37

chadhietala mentioned this pull request Jun 10, 2016

Glimmer Pre-flight Checklist #13644

Closed

asakusuma mentioned this pull request Jun 10, 2016

Add location parameters to builder functions glimmerjs/glimmer-vm#182

Merged

asakusuma force-pushed the input-dynamic branch from 4cb03d1 to 2d794a3 Compare June 13, 2016 17:54

krisselden added the Glimmer2 label Jun 14, 2016

asakusuma force-pushed the input-dynamic branch from 2d794a3 to e90809a Compare June 16, 2016 21:49

asakusuma force-pushed the input-dynamic branch 2 times, most recently from 15e576f to 05f25dd Compare June 20, 2016 21:26

asakusuma force-pushed the input-dynamic branch from 05f25dd to 73b9a01 Compare June 21, 2016 19:54

tomdale mentioned this pull request Jun 21, 2016

Include TypeScript source in npm package glimmerjs/glimmer-vm#190

Merged

asakusuma force-pushed the input-dynamic branch from 73b9a01 to 537217a Compare June 21, 2016 21:02

asakusuma force-pushed the input-dynamic branch from 537217a to cf9bcbd Compare June 21, 2016 21:38

asakusuma mentioned this pull request Jun 21, 2016

Fix how input handles initial null attributes glimmerjs/glimmer-vm#195

Merged

2 tasks

asakusuma force-pushed the input-dynamic branch from cf9bcbd to a28aa47 Compare June 22, 2016 00:17

[Glimmer2] Implement input

22e1b13

asakusuma force-pushed the input-dynamic branch from a28aa47 to 22e1b13 Compare June 22, 2016 00:58

rwjblue merged commit eccb363 into emberjs:master Jun 22, 2016

fivetanley pushed a commit to fivetanley/ember.js that referenced this pull request Jul 17, 2016

[glimmer2] enable tests fixed by emberjs#13615

d665c55

refs emberjs#13644

chancancode pushed a commit that referenced this pull request Jul 17, 2016

[glimmer2] enable tests fixed by #13615

8263842

refs #13644

toddjordan pushed a commit to toddjordan/ember.js that referenced this pull request Sep 9, 2016

[glimmer2] enable tests fixed by emberjs#13615

ee5e8d1

refs emberjs#13644

[Glimmer2] Implement input #13615

[Glimmer2] Implement input #13615

Conversation

asakusuma commented Jun 7, 2016

chadhietala Jun 7, 2016

Choose a reason for hiding this comment

asakusuma Jun 7, 2016

Choose a reason for hiding this comment

asakusuma commented Jun 7, 2016

asakusuma commented Jun 7, 2016

asakusuma commented Jun 8, 2016

rwjblue commented Jun 8, 2016

rwjblue Jun 8, 2016

Choose a reason for hiding this comment

rwjblue Jun 8, 2016

Choose a reason for hiding this comment

asakusuma Jun 8, 2016

Choose a reason for hiding this comment

rwjblue Jun 8, 2016

Choose a reason for hiding this comment

asakusuma Jun 8, 2016 • edited Loading

Choose a reason for hiding this comment

rwjblue commented Jun 8, 2016

asakusuma commented Jun 8, 2016

asakusuma commented Jun 8, 2016

chadhietala commented Jun 10, 2016

homu commented Jun 11, 2016

krisselden commented Jun 14, 2016

asakusuma commented Jun 14, 2016

homu commented Jun 16, 2016

asakusuma commented Jun 17, 2016

rwjblue commented Jun 17, 2016

asakusuma commented Jun 17, 2016

homu commented Jun 18, 2016

jdalton commented Jun 20, 2016

rwjblue commented Jun 20, 2016

jdalton commented Jun 20, 2016 • edited Loading

asakusuma commented Jun 20, 2016

jdalton commented Jun 21, 2016 • edited Loading

homu commented Jun 21, 2016

asakusuma commented Jun 21, 2016

tomdale commented Jun 21, 2016

asakusuma commented Jun 21, 2016

asakusuma commented Jun 22, 2016

asakusuma commented Jun 22, 2016

rwjblue commented Jun 22, 2016

rwjblue commented Jun 22, 2016

asakusuma Jun 8, 2016 •

edited

Loading

jdalton commented Jun 20, 2016 •

edited

Loading

jdalton commented Jun 21, 2016 •

edited

Loading