end sentences with a .

this half-sentence just repeats what the first paragraph already said?

The docs say enableInteractiveSelection is true by default. What does it mean if it is null? (Are we just missing an assert in the constructor that this cannot be null?)

it default to null, I have though about default it to true in constructor and just do bool get selectionEnabled => enableInteractiveSelection

but that will require assert(enableInteractiveSelection!= null) which will change api behavior, previously we allow user pass in null.

I think changing it to include the assert will be fine. That we don't have it is more like a bug...

we dont have that because we default it based on obscureText. I will fix this, do you think this will be a breaking change?

why not use macros for these as well?

It has a different default from editable text. same for copy

Is this the right way to get the full length of the text? @GaryQian may know.

Generally, this will indeed produce the length of the text content of the text object. However, to properly and explicitly handle things like PlaceholderSpan/WidgetSpans, you may have to look deeper.

toPlainText will return the concatenated contents for TextSpan. When encountering PlaceholderSpans, the placeholder span will be ignored unless includePlaceholders is true, in which case a 0xFFFC character of length 1 will be included. Whether or not you want this replacement character is a policy choice. In addition, if another text widget is nested using a WidgetSpan, the length of the nested contents will not be included.

This looks like it is for obscuring text in a password like environment, which I expect would not be using PlaceholderSpans. It is hard to say what developers would expect if one were to, say, embed all of flutter gallery inline within a password field. How many character lengths is a gallery equivalent to?

Just setting the includePlaceholders flag to true seems to me like a reasonable compromise, where all placeholders/inline widgets will be represented as one character.

the includePlaceholders is defaulted to true. As of now, there is no way to use any other inlinespan other than textspan in RenderEditable. There is no plan to add it in until we figure out how do we serialize them. We only need to worry about text span for now.

Edit, I was wrong, user can do anything if they use RenderEditable directly... but still we don't know how should we deal with non textspan. I believe let includePlaceholders defaults to true will be good enough.

half-sentence?

These already have a lot of options... would it maybe make sense to give them a single toolbarOptions property that encapsulates all of these and nicely groups them?

nit: space after :

nit: space after :

Maybe include a sentence indicating what the toolbar is used for?

Can this (and the others) just be this.copy = false?

And then just assert that they are not null?

The comment says there should be 3, why did that change?

it used to be copy, paste, and cut.
after this change it by default can only paste

same here.

move this to the line above (or indent).

same

Why a double tap? Shouldn't a single tap be enough deselect?

Can you just use longPressAt (https://master-api.flutter.dev/flutter/flutter_test/WidgetController/longPressAt.html) here and elsewhere where you needed a long press?

Did you think about asserting or throwing an exception if cut/paste is true and readOnly is true? I'm not sure which is better, just wondering if you thought about it.

It seems weird to me, but asserting the property of a toolbarOptions is not a constant expression. This prevent me from doing the check in constant constructor. I might be able to do the check in build, but do you think it is a good approach?
As of whether we should enforce this, I think we shouldn't because It probably will be very common to switch between readonly and non-readonly. It feel weird to me that if I switch to read only i have to explicit disable the option.
But I do want to enforce it in selectable text though but was thinking whether worth it to make the constructor non-const or checking it outside the constructor.

Oh hmm I thought you could add an assert in the initializer list while still being const. If not then maybe it's not worth it.

Maybe write this "Toolbar" instead of "Tool bar" (here and below)? I think putting a space in between implies that the B should be capitalized when camel cased ("ToolBar"), but it's not.

This is a definite nitpick though, totally up to you :)

Can you explain this section here? Doesn't "isCollapsed true" mean that the text is not selected, or am I reading this wrong?

if it is selectable, a long press will cause it to select the whole word and the selection will not be collapsed.

I inverted this test and the test above because we are inverted the selection behavior of password textfield.

Are the comments reversed then? Here it says "Long press does select text." and then it asserts that isCollapsed is true, meaning that there is no selection. Above it says "Long press doesn't select anything." and then asserts isCollapsed is false. Or maybe I'm just confusing myself haha.

oh I forgot to update the comment, thanks for catching it