Issue 59220044: make ref bindable

rafaelw

ptal. note that having Template capture bind here is only necessary in cases where MutationObservers ...

11 years, 2 months ago (2014-02-06 23:31:32 UTC) #1

rafaelw

Committed patchset #3 manually as r2799341 (presubmit successful).

11 years, 2 months ago (2014-02-06 23:56:59 UTC) #2

arv

https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js File src/TemplateBinding.js (right): https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js#newcode483 src/TemplateBinding.js:483: templateObserver.observe(this, { attributes: true, Why are we doing this ...

11 years, 2 months ago (2014-02-07 18:43:42 UTC) #3

rafaelw

https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js File src/TemplateBinding.js (right): https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js#newcode483 src/TemplateBinding.js:483: templateObserver.observe(this, { attributes: true, It's an interesting point. Do ...

11 years, 2 months ago (2014-02-09 00:33:16 UTC) #4

arv

I guess I just don't understand the motivation for treating the ref attribute this way. ...

11 years, 2 months ago (2014-02-10 16:13:54 UTC) #5

rafaelw1

I willing to be convinced. My answer is that ref is useful as a bindable ...

11 years, 2 months ago (2014-02-11 22:02:09 UTC) #6

John Messerly

Really good question :) So with "ref" the idea seems to be that it's a ...

11 years, 2 months ago (2014-02-11 23:10:35 UTC) #7

Really good question :)

So with "ref" the idea seems to be that it's a DOM property of
HTMLTemplateElement, so it makes sense to reflect the property to the
attribute and vice versa. Likewise, it makes sense for it to be bindable.

Currently "bind" and "repeat" are not properties of HTMLTemplateElement. In
some sense "bind" corresponds to "model". They aren't really interpreted
except as part of <template> instantiation.

IMHO, it seems like generally a good thing if DOM attributes have
corresponding properties, and most of them do. At that point it would be
nice if "bind" and "repeat" worked the same as "ref": the property responds
to attribute changes, and is data bindable via Node.bind. Whether it is
useful or not, I'm not sure. I think the main benefit would be general
consistency (especially if "bind" and "model" could be reconciled) but if
no one has asked for it, maybe not high priority.

On Tue, Feb 11, 2014 at 2:02 PM, Rafael Weinstein <rafaelw@google.com>wrote:

> I willing to be convinced. My answer is that ref is useful as a bindable
> attribute. It's not to hard to imagine switching what is getting stamped
> based on something dynamic. Binding to the bind/repeat/if seems less
> obviously useful (at least without asking questions why you didn't do it
> differently).
>
> John?
>
>
> On Mon, Feb 10, 2014 at 8:13 AM, Erik Arvidsson <arv@chromium.org> wrote:
>
>> I guess I just don't understand the motivation for treating the ref
>> attribute this way. Why does this one warrant two way binding for the
>> attribute. We don't allow it for any other attribute .
>>
>> On Sat, Feb 8, 2014 at 7:33 PM,  <rafaelw@chromium.org> wrote:
>> >
>> >
>> https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js
>> > File src/TemplateBinding.js (right):
>> >
>> >
>>
https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js#new...
>> > src/TemplateBinding.js:483: templateObserver.observe(this, { attributes:
>> > true,
>> > It's an interesting point. Do you think that the attribute values of if,
>> > bind, & repeat should be observed and react dynamically as well? ref is
>> > somewhat different in that it is bindable, where-as there's no way to
>> > bind to the attribute value of bind, if & repeat.
>> >
>> >
>> > On 2014/02/07 18:43:42, arv wrote:
>> >>
>> >> Why are we doing this for ref but not for bind and repeat?
>> >
>> >
>> > https://codereview.appspot.com/59220044/
>>
>>
>>
>> --
>> erik
>>
>
>

Sign in to reply to this message.

John Messerly

(oops, meant to CC Pete, he is pretty interested in this topic too) On Tue, ...

11 years, 2 months ago (2014-02-11 23:11:29 UTC) #8

(oops, meant to CC Pete, he is pretty interested in this topic too)


On Tue, Feb 11, 2014 at 3:09 PM, John Messerly <jmesserly@google.com> wrote:

> Really good question :)
>
> So with "ref" the idea seems to be that it's a DOM property of
> HTMLTemplateElement, so it makes sense to reflect the property to the
> attribute and vice versa. Likewise, it makes sense for it to be bindable.
>
> Currently "bind" and "repeat" are not properties of HTMLTemplateElement.
> In some sense "bind" corresponds to "model". They aren't really interpreted
> except as part of <template> instantiation.
>
> IMHO, it seems like generally a good thing if DOM attributes have
> corresponding properties, and most of them do. At that point it would be
> nice if "bind" and "repeat" worked the same as "ref": the property responds
> to attribute changes, and is data bindable via Node.bind. Whether it is
> useful or not, I'm not sure. I think the main benefit would be general
> consistency (especially if "bind" and "model" could be reconciled) but if
> no one has asked for it, maybe not high priority.
>
>
> On Tue, Feb 11, 2014 at 2:02 PM, Rafael Weinstein <rafaelw@google.com>wrote:
>
>> I willing to be convinced. My answer is that ref is useful as a bindable
>> attribute. It's not to hard to imagine switching what is getting stamped
>> based on something dynamic. Binding to the bind/repeat/if seems less
>> obviously useful (at least without asking questions why you didn't do it
>> differently).
>>
>> John?
>>
>>
>> On Mon, Feb 10, 2014 at 8:13 AM, Erik Arvidsson <arv@chromium.org> wrote:
>>
>>> I guess I just don't understand the motivation for treating the ref
>>> attribute this way. Why does this one warrant two way binding for the
>>> attribute. We don't allow it for any other attribute .
>>>
>>> On Sat, Feb 8, 2014 at 7:33 PM,  <rafaelw@chromium.org> wrote:
>>> >
>>> >
>>> https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js
>>> > File src/TemplateBinding.js (right):
>>> >
>>> >
>>>
https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js#new...
>>> > src/TemplateBinding.js:483: templateObserver.observe(this, {
>>> attributes:
>>> > true,
>>> > It's an interesting point. Do you think that the attribute values of
>>> if,
>>> > bind, & repeat should be observed and react dynamically as well? ref is
>>> > somewhat different in that it is bindable, where-as there's no way to
>>> > bind to the attribute value of bind, if & repeat.
>>> >
>>> >
>>> > On 2014/02/07 18:43:42, arv wrote:
>>> >>
>>> >> Why are we doing this for ref but not for bind and repeat?
>>> >
>>> >
>>> > https://codereview.appspot.com/59220044/
>>>
>>>
>>>
>>> --
>>> erik
>>>
>>
>>
>

Sign in to reply to this message.

John Messerly

https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js File src/TemplateBinding.js (right): https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js#newcode443 src/TemplateBinding.js:443: return Element.prototype.bind.call(this, name, value, oneTime); here's a question, can ...

11 years, 2 months ago (2014-02-11 23:14:51 UTC) #9

blois

Not to rathole, but my initial feedback on ref was that I expected it to ...

11 years, 2 months ago (2014-02-11 23:32:37 UTC) #10

Not to rathole, but my initial feedback on ref was that I expected it to
take an HTMLTemplateElement instance rather than an ID. The issue that I
was running into was that I had a complex custom element hierarchy and
wanted to pass an item renderer (template) from one location to a deeply
nested location. For 'repeat', I was expecting a property which is the
array to iterate over. That said, with complex controls and virtualization
I've mostly moved away from template repeat- but it would be nice if the
APIs could be consistent.


On Tue, Feb 11, 2014 at 3:10 PM, John Messerly <jmesserly@google.com> wrote:

> (oops, meant to CC Pete, he is pretty interested in this topic too)
>
>
> On Tue, Feb 11, 2014 at 3:09 PM, John Messerly <jmesserly@google.com>wrote:
>
>> Really good question :)
>>
>> So with "ref" the idea seems to be that it's a DOM property of
>> HTMLTemplateElement, so it makes sense to reflect the property to the
>> attribute and vice versa. Likewise, it makes sense for it to be bindable.
>>
>> Currently "bind" and "repeat" are not properties of HTMLTemplateElement.
>> In some sense "bind" corresponds to "model". They aren't really interpreted
>> except as part of <template> instantiation.
>>
>> IMHO, it seems like generally a good thing if DOM attributes have
>> corresponding properties, and most of them do. At that point it would be
>> nice if "bind" and "repeat" worked the same as "ref": the property responds
>> to attribute changes, and is data bindable via Node.bind. Whether it is
>> useful or not, I'm not sure. I think the main benefit would be general
>> consistency (especially if "bind" and "model" could be reconciled) but if
>> no one has asked for it, maybe not high priority.
>>
>>
>> On Tue, Feb 11, 2014 at 2:02 PM, Rafael Weinstein <rafaelw@google.com>wrote:
>>
>>> I willing to be convinced. My answer is that ref is useful as a bindable
>>> attribute. It's not to hard to imagine switching what is getting stamped
>>> based on something dynamic. Binding to the bind/repeat/if seems less
>>> obviously useful (at least without asking questions why you didn't do it
>>> differently).
>>>
>>> John?
>>>
>>>
>>> On Mon, Feb 10, 2014 at 8:13 AM, Erik Arvidsson <arv@chromium.org>wrote:
>>>
>>>> I guess I just don't understand the motivation for treating the ref
>>>> attribute this way. Why does this one warrant two way binding for the
>>>> attribute. We don't allow it for any other attribute .
>>>>
>>>> On Sat, Feb 8, 2014 at 7:33 PM,  <rafaelw@chromium.org> wrote:
>>>> >
>>>> >
>>>> https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js
>>>> > File src/TemplateBinding.js (right):
>>>> >
>>>> >
>>>>
https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js#new...
>>>> > src/TemplateBinding.js:483: templateObserver.observe(this, {
>>>> attributes:
>>>> > true,
>>>> > It's an interesting point. Do you think that the attribute values of
>>>> if,
>>>> > bind, & repeat should be observed and react dynamically as well? ref
>>>> is
>>>> > somewhat different in that it is bindable, where-as there's no way to
>>>> > bind to the attribute value of bind, if & repeat.
>>>> >
>>>> >
>>>> > On 2014/02/07 18:43:42, arv wrote:
>>>> >>
>>>> >> Why are we doing this for ref but not for bind and repeat?
>>>> >
>>>> >
>>>> > https://codereview.appspot.com/59220044/
>>>>
>>>>
>>>>
>>>> --
>>>> erik
>>>>
>>>
>>>
>>
>

Sign in to reply to this message.

rafaelw1

11 years, 2 months ago (2014-02-25 21:17:50 UTC) #11

I'm willing to consider allowing the ref binding to accept both a string
which is treated as an id reference, or an actual template element
reference.

I'm interested in hearing specific use cases for the later.


On Tue, Feb 11, 2014 at 3:32 PM, Pete Blois <blois@google.com> wrote:

> Not to rathole, but my initial feedback on ref was that I expected it to
> take an HTMLTemplateElement instance rather than an ID. The issue that I
> was running into was that I had a complex custom element hierarchy and
> wanted to pass an item renderer (template) from one location to a deeply
> nested location. For 'repeat', I was expecting a property which is the
> array to iterate over. That said, with complex controls and virtualization
> I've mostly moved away from template repeat- but it would be nice if the
> APIs could be consistent.
>
>
> On Tue, Feb 11, 2014 at 3:10 PM, John Messerly <jmesserly@google.com>wrote:
>
>> (oops, meant to CC Pete, he is pretty interested in this topic too)
>>
>>
>> On Tue, Feb 11, 2014 at 3:09 PM, John Messerly <jmesserly@google.com>wrote:
>>
>>> Really good question :)
>>>
>>> So with "ref" the idea seems to be that it's a DOM property of
>>> HTMLTemplateElement, so it makes sense to reflect the property to the
>>> attribute and vice versa. Likewise, it makes sense for it to be bindable.
>>>
>>> Currently "bind" and "repeat" are not properties of HTMLTemplateElement.
>>> In some sense "bind" corresponds to "model". They aren't really interpreted
>>> except as part of <template> instantiation.
>>>
>>> IMHO, it seems like generally a good thing if DOM attributes have
>>> corresponding properties, and most of them do. At that point it would be
>>> nice if "bind" and "repeat" worked the same as "ref": the property responds
>>> to attribute changes, and is data bindable via Node.bind. Whether it is
>>> useful or not, I'm not sure. I think the main benefit would be general
>>> consistency (especially if "bind" and "model" could be reconciled) but if
>>> no one has asked for it, maybe not high priority.
>>>
>>>
>>> On Tue, Feb 11, 2014 at 2:02 PM, Rafael Weinstein
<rafaelw@google.com>wrote:
>>>
>>>> I willing to be convinced. My answer is that ref is useful as a
>>>> bindable attribute. It's not to hard to imagine switching what is getting
>>>> stamped based on something dynamic. Binding to the bind/repeat/if seems
>>>> less obviously useful (at least without asking questions why you didn't do
>>>> it differently).
>>>>
>>>> John?
>>>>
>>>>
>>>> On Mon, Feb 10, 2014 at 8:13 AM, Erik Arvidsson <arv@chromium.org>wrote:
>>>>
>>>>> I guess I just don't understand the motivation for treating the ref
>>>>> attribute this way. Why does this one warrant two way binding for the
>>>>> attribute. We don't allow it for any other attribute .
>>>>>
>>>>> On Sat, Feb 8, 2014 at 7:33 PM,  <rafaelw@chromium.org> wrote:
>>>>> >
>>>>> >
>>>>> https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js
>>>>> > File src/TemplateBinding.js (right):
>>>>> >
>>>>> >
>>>>>
https://codereview.appspot.com/59220044/diff/40001/src/TemplateBinding.js#new...
>>>>> > src/TemplateBinding.js:483: templateObserver.observe(this, {
>>>>> attributes:
>>>>> > true,
>>>>> > It's an interesting point. Do you think that the attribute values of
>>>>> if,
>>>>> > bind, & repeat should be observed and react dynamically as well? ref
>>>>> is
>>>>> > somewhat different in that it is bindable, where-as there's no way to
>>>>> > bind to the attribute value of bind, if & repeat.
>>>>> >
>>>>> >
>>>>> > On 2014/02/07 18:43:42, arv wrote:
>>>>> >>
>>>>> >> Why are we doing this for ref but not for bind and repeat?
>>>>> >
>>>>> >
>>>>> > https://codereview.appspot.com/59220044/
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> erik
>>>>>
>>>>
>>>>
>>>
>>
>

Sign in to reply to this message.

Issue 59220044: make ref bindable (Closed)

Description

Patch Set 1 #

Patch Set 2 : moar #

Patch Set 3 : moar #

Messages