Crashes extension host when parsing emoji, for example 🇮🇳

[rhnvrm]

Version information:
VSCode: 1.34
SpellRight: 3.0.38

To reproduce, copy and paste this emoji (🇮🇳) in a markdown file with spellright enabled.

Probably related to (#223)

[danmackinlay]

@rhnvrm What platform are you on?
I'm seeing a spellRight crash on emoji, but not on VS Code macOS, only VS Code Ubuntu.

This markdown fragment causes spellright to crash:

### Android

🤷‍♂️

It also renders badly in VS Code, although that is not the fault of spellright.

$ code --status --show-versions                                                                          
Version:          Code 1.36.1 (2213894ea0415ee8c85c5eea0d0ff81ecc191529, 2019-07-08T22:55:08.091Z)
OS Version:       Linux x64 4.15.0-54-generic
...
$ code  --status --show-versions --list-extensions
ban.spellright@3.0.46
...

I get an actual crashlog sometimes from the crashing extension host. I will attach it here next time it occurs.

[rhnvrm]

My current machine is using:
OS Version: Linux x64 4.15.0-51-generic

I had tested this on Arch Linux (VS Codium) and Ubuntu (VS Code) both.

[danmackinlay]

Now that I know it's Linux VS code with emoji, at least I'll be able to test some alternative distros tomorrow in the office.

Anyway, I'm having trouble getting the crash report working. For now, I'll confirm I'm getting the SIGSEGV reported in a function leftcommonsubstring somewhere in the hunspell adapter.

I also get a spellright hanging a few times; the CPU profiles from that I DO have, in /tmp/exthost-*.cpuprofile, but that's presumably a different issue.

[howthebodyworks]

This is @danmackinlay on my work github account.

The problem also arises on Kubuntu 16.04. Here are some screenshot error logs from apport (which I should work out how to export).

Once again, triggered only by documents with emoji. (On Kubuntu, support for these is even worse)

[jacksteamdev]

For me, these emojis: 🧐, 🤩. Both pasting and opening a file with these chars causes a crash.

FWIW, these are both relatively new emojis, both added in 2017 in Unicode 10.0 and Emoji 5.0.

About:

Version: 1.36.1
Commit: 2213894ea0415ee8c85c5eea0d0ff81ecc191529
Date: 2019-07-08T22:55:08.091Z
Electron: 4.2.5
Chrome: 69.0.3497.128
Node.js: 10.11.0
V8: 6.9.427.31-electron.0
OS: Linux x64 5.0.0-20-generic

[jgehrcke]

@bartosz-antosik do you plan to debug and address this?

[laymonage]

For me, the 🇪🇺 emoji causes this (as well as the emojis mentioned above). It would be really nice if this is solved.

[thetooth]

I am in the middle of writing an extremely important whitepaper in restructuredtext and need to use the 🤔 emoji but am encountering this issue. Has there been any progress with this? It also happens that the rst extension hooks the backspace key, and since everything crashes together, I am unable to remove the offending glyph without using another editor. The professor will be furious if I turn this in late again.

[brunetton]

Same here. It took me some time to understand that 🤔 emoji caused a crash of the extension