bug#34969: 26.2; `emacs -Q` unable to copy \200's eight-bit detail buffer

[Van L <van@scratch.space>]

> Eli Zaretskii writes:
> 
>> From: Van L <van@scratch.space>
>> Date: Tue, 26 Mar 2019 10:46:12 +1100
>> Cc: 34969@debbugs.gnu.org
>> 
>> Copy and paste within the one Emacs is being interfered with if what you suggest is true.
> 
> When you paste, and there's stuff in the clipboard, Emacs checks
> whether Emacs itself is the owner of that stuff.  If it is, we yank
> from the kill ring instead of accessing the clipboard, because yanking
> is both faster and doesn't need any decoding.
> 
> So I'm guessing that some software on your system takes ownership of
> the text in the clipboard, and Emacs then accesses that via the
> window-system selections, and fails to decode raw bytes due to
> character encoding issues.

Could be.

>>> Section A just gives a single sequence, but it doesn't tell how to do
>>> that several times, i.e. what exactly is repeated N times.
>> 
>> a. run Emacs
>> b. goto *scratch* buffer
>> c. create new frame
>> d. lookup keybindings for *scratch* buffer
>> e. find `self`
>> f. for the two variants of \200 copy and paste details buffer to *scratch*
>> 
>> Repeat the above a-f steps 7 times as listed in the bug-report for those platforms detailed.
> 
> We are miscommunicating, I think.  Each one of the 7 instances in your
> report says something like this, for example:
> 
>> : 4
>> : [has problem the second time]
>> : GNU Emacs 26.2 (build 1, x86_64-apple-darwin15.6.0, X toolkit)
>> : of 2019-03-21
> 
> I'm asking what exactly does "second time" mean here.  Did you start
> this version of Emacs twice, or did you start it only once and
> performed the steps b-f above twice one after the other in the same
> Emacs session?

Emacs was started 7-times and a-f steps performed in each once.


>>>>> What happens if you set interprogram-cut-function and
>>>>> interprogram-paste-function to nil?
>>>> 
>>>> Will do the next time I get a chance.
>>> 
>>> If I'm right, the problem will disappear.
>> 
>> I am inside the one Emacs's copy and paste, and not going out of the Emacs.
> 
> I don't think I understand what this alludes to.  Is it a response to
> my sentence that starts with "If I'm right"?

I was focusing on `interprogram-cut/paste` and drawing from that Emacs and other programs-inter-cutting-and-pasting with the clipboard. I wanted to make certain to you that no other cut/paste operation happened outside of Emacs for the duration of a-f steps.

The response to "If I'm right" can't be confirmed until I clean slate test with the same conditions observed.

>>>>> (And why do your symlink targets include \200 characters?)
>>>> 
>>>> I used `C-x 8 RET horizontal ellipsis` the email program was unable to process.
>>>> The \200 was what the email program used as substitute.
>>> 
>>> Strange email program: \200 is an unassigned codepoint.
>> 
>> That was the emacs-bug-report mechanism in Emacs that did that.
> 
> If you sent the report via mailclient, then this is probably the same
> problem which causes the issue with pasting that we are discussing.

The mailclient was Emacs's. It offered a menu for processing the problematic characters.

The pasting problem we are discussing is failure to paste what was selected and insertion of an ealier copy from Emacs to the clipboard.

Hope that helps.