Re: [PATCH v5] Enable xwidgets on macOS

[조성빈]

> 2019. 7. 30. 오전 5:26, Alan Third <address@hidden> 작성:
> 
> On Fri, Jul 26, 2019 at 03:51:11AM +0900, 조성빈 wrote:
>> Hello, I have prepared three patches to apply on master.
>> Can anyone review these?
> 
> This error message is extremely annoying:
> 
> You can’t share an xwidget (webkit2) among windows. [47 times]
> mwheel-scroll: Beginning of buffer
> You can’t share an xwidget (webkit2) among windows. [87 times]
> completing-read-default: Command attempted to use minibuffer while in 
> minibuffer
> You can’t share an xwidget (webkit2) among windows. [4 times]
> Quit [4 times]
> You can’t share an xwidget (webkit2) among windows. [121 times]
> 
> It blocks the minibuffer so it’s impossible to see what you’re typing
> when you do M-x or C-x b and so on. Is it possible to limit it to only
> when an actual xwidget buffer is selected?

Wow, I found I just got used to that before I started to maintain the patch and 
never even thought to fix that. I remember this was annoying to hell to me too 
:-(. I'll try to find a solution.

> Strange things also happen if you have the same xwidget buffer
> displayed in two windows and then resize the one that doesn’t display
> the xwidget. It looks as though the wrong window is being used to
> calculate the xwidget’s size.

Hmm, maybe this is because the portion of xwidget in the window isn't redrawn? 
I'm not sure how I should tackle the issue though. :-(

> The big explanatory comment at the top of nsxwidget.m could do with
> being rewritten. It’s got a few typos and I find hard to understand. I
> think I get the gist of it, but I’m not 100% sure.

Ok, I'll try to clarify it.

> Is this something we’re going to have to update over time?
> 
>      self.customUserAgent =
>        @"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_6)"
>        @" AppleWebKit/603.3.8 (KHTML, like Gecko)"
>        @" Version/11.0.1 Safari/603.3.8";

I removed this part.

> Anyway, the code contains two calls to
> evaluateJavaScript:completionHandler: which is a problem as we’re
> banned from using ObjC blocks, and the completion handler is an ObjC
> block. Unless GCC supports them now, which I doubt.

Does that mean this patch can never get merged until gcc supports blocks?

> I don’t see any alternatives in the Apple documentation.