Notepad++ v8.6 Release Candidate

PeterJones

@donho said in Notepad++ v8.6 Release Candidate:

Is anyone here can reproduce it?

Yes, I can reproduce (though at first I couldn’t) – maybe like me, @donho did a 4x1 column selection instead of 4x0 column selection:

Step 1: column selection, showing it’s one character wide:

Step 2: hit Del/Delete key: it deletes the whole NBSP, without leaving

Confirm: View > Show Symbol is showing all characters, so it’s not just hiding something:

Original Conclusion: thought I could not reproduce. But as I was re-reading @mkupper’s post, I wondered if I misinterpreted.

Undo, so it’s back to the original

Then Step 1: do a 4x0 zero-width column selection

Step 2: Del/Delete =>

Confirmed the bug report. With a 0-width column selection, hitting Del before NBSP broke it into its constituent bytes and just deleted one byte rather than one character.

Tried with some accented characters (2 bytes each) and then some smilies (with 3-4 bytes each):

Before:

After hitting Del once:

So it’s not just NBSP. With a 0-width column selection, hitting Del when the next character is a multi-byte UTF-8 character causes it to delete only the first byte rather than the whole character.

PeterJones

@donho said in Notepad++ v8.6 Release Candidate:

What’s the Hex value of NBSB? So I can try to reproduce it.

@mkupper’s NBSB was just a typo for NBSP; see the right value in his screenshot

NBSP is the non-breaking space, at U+00A0 , which is UTF-8 encoded as the two bytes c2a0.

donho

@PeterJones @mkupper
Please check the fix:
https://github.com/notepad-plus-plus/notepad-plus-plus/pull/14384

PeterJones

@donho ,

I first tried with the “Notepad++.MSVC.x64.Debug” artifact: I loaded my example file from yesterday, did the zero-width selection, and hit Delete, and that caused an MSVC assertion error:

… But as I was writing this reply, I realized that I probably don’t have everything set up right for DEBUG builds, so I thought I would try the RELEASE build:

When I run with the “Notepad++.MSVC.x64.Release” artifact, the zero-width column selection followed by Del works as expected on the multi-byte UTF-8 characters. So the fix looks good to me. But you should probably wait for @mkupper to confirm as well, because he might have a different sequence, which may or may not behave the same as my experiment.

Alan Kilborn

@PeterJones said in Notepad++ v8.6 Release Candidate:

I realized that I probably don’t have everything set up right for DEBUG builds

There is no special setup for running a Debug build, so you are set up correctly. :-)

The problem is: Someone at some time wrote some poor assertions, and now they just get in the way of things. OR…the assertions were good at the time written, but now other things have changed, and no one maintained the assertions, so now they aren’t so good. :-(

donho

@PeterJones said in Notepad++ v8.6 Release Candidate:

I first tried with the “Notepad++.MSVC.x64.Debug” artifact: I loaded my example file from yesterday, did the zero-width selection, and hit Delete, and that caused an MSVC assertion error:

Could you share your example file here please?

PeterJones

@donho said in Notepad++ v8.6 Release Candidate:

Could you share your example file here please?

768 × 1024
768 × 1024
768 × 1024
768 × 1024
U+C0	À	À
U+C1	Á	Á
U+C2	Â	Â
U+C3	Ã	Ã
U+C4	Ä	Ä
U+C5	Å	Å
* ☺ `U+2640`       SMILE        ☺
* ☹ `U+2639`       FROWN       ☹
* 😃 `U+1F603`      OPEN        😃
* 😉 `U+1F609`      WINK        😉
* 😐 `U+1F610`      NEUTRAL     😐
* 😜 `U+1F61C`      TongueOut   😜
* 😲 `U+1F632`      ASTONISHED  😲

update: the tabs and NBSP didn’t come through, so here it is base64-encoded using the MIME Tools plugin:

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

donho

@PeterJones said in Notepad++ v8.6 Release Candidate:

update: the tabs and NBSP didn’t come through, so here it is base64-encoded using the MIME Tools plugin:

Hmm… using debug built locally binary, with 0 length column mode selection in front of all special chars then DEL - no assertion.

Could you provide the accurate instructions to reproduce the assertion please?

PeterJones

@donho ,

I found the problem on that. When I got the assertion error, I was still using a corrupted version of my testfile that had some of the NBSP as a single xA0 instead.

I can also reproduce the error by pasting in this base64 and decoding it:

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

So the error is unrelated to the column-delete, and instead related to the DEBUG-build’s handling of invalid UTF-8 sequences in certain unusual situations. (I don’t think it’s critical that a debug build has difficuly with invalid UTF-8. I verified that the v8.6-RC build and the RELEASE build on this PR don’t do anything unpleasant under the same circumstances. Unless you want to make the DEBUG build more robust against invalid UTF-8, I’d probably recommend not worrying about it.)

donho

@PeterJones

Thank you for the detail explanation.
I will do a RC2 immediately.

donho

Notepad++ v8.6 Release Candidate 2 is available here:
https://community.notepad-plus-plus.org/topic/25158/notepad-v8-6-release-candidate-2

The regression of DEL is fixed in RC2.

@xomx
https://github.com/notepad-plus-plus/wingup/pull/48 is included.