neovim stacktrace when there is no history

[dsummersl]

"__Mundo__" [New File]
Error detected while processing function mundo#MundoToggle[1]..<SNR>290_MundoToggle[11]..<SNR>290_MundoOpen[31]..<SNR>290_MundoPython[2]..provider#python3#Call:
line   18:
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/Users/danesummers/environment/vim/bundle/gundo.vim/autoload/mundo.py", line 107, in MundoRenderGraph
    nodesData
  File "/Users/danesummers/environment/vim/bundle/gundo.vim/autoload/mundo/graphlog.py", line 218, in generate
    util._undo_to(current)
  File "/Users/danesummers/environment/vim/bundle/gundo.vim/autoload/mundo/util.py", line 34, in _undo_to
    vim().command('silent undo %d' % int(n))
  File "/usr/local/lib/python3.5/site-packages/neovim/api/nvim.py", line 218, in command
    return self.request('nvim_command', string, **kwargs)
  File "/usr/local/lib/python3.5/site-packages/neovim/api/nvim.py", line 131, in request
    res = self._session.request(name, *args, **kwargs)
  File "/usr/local/lib/python3.5/site-packages/neovim/msgpack_rpc/session.py", line 98, in request
    raise self.error_wrapper(err)
neovim.api.nvim.NvimError: b'Vim(undo):E830: Undo number 1 not found'
Error detected while processing function mundo#MundoToggle[1]..<SNR>290_MundoToggle[11]..<SNR>290_MundoOpen[31]..<SNR>290_MundoPython:
line    2:
E171: Missing :endif
Error detected while processing function mundo#MundoToggle[1]..<SNR>290_MundoToggle:
line   11:
E171: Missing :endif

[simnalamburt]

Oops it's a big problem. I didn't know this since I do not use neovim.

Since all vim plugin should work fine also in neovim, probably it's a bug of neovim. Maybe we can do both report this problem to the neovim and workaround this bug in vim-mundo.

[wsdjeg]

@dsummersl I use neovim, and it works well for me , what is your reproduce step?

[dsummersl]

Struggling to figure out how to reproduce. It has happened to me a few times in the past (I presume, I've seen similar stack traces but I didn't copy the traces the last times, so I can't be 100% sure). What I know thus far:

• It happens for files with no history.
• Normally a file with no history would have 1 undo. :echo undotree() there should normally only have 1 undo entry with a level of 0. The stack trace here shows that mundo is attempting to go to level 1.

[wsdjeg]

I just create a new file by echo "" > test/test.txt, then use nvim open such file, and echo undotree(), open moundo. here is the pic.

[dsummersl]

Yeah, I've tried that too...and using it on a read only file too. I'm not sure how to trigger it. Again, this doesn't happen 'normally' - but there is a stack trace - so its happened! :)

I think it has happened with file I edit under version control. If I work on branch A for a file, and then switch to branch B(so that whatever undotree() state that was valid for the file on branch A gets thrown away by vim, b/c the file has changed - I guess this is only true for files that are different on the two branches)

[dsummersl]

I'll try to make it happen today.

[vimeitor]

I can reproduce it fairly consistently. An example:

Open a file
Delete a char
Save
Undo
Save
Run "set undoreload=0 | edit"
Delete a char
Save
Undo
Save
Run "set undoreload=0 | edit"
Run mundo

set undoreload=0 | edit basically discards the undo history for the current file. Let me know if you can reproduce the problem :)

[dsummersl]

thanks @vimeitor I'll give that a shot

[vimeitor]

I forgot to say that this is not something that is exclusive of neovim, as I can reproduce it in vim8, as well.

[dsummersl]

@vimeitor I've merged in a fix that I believe corrects the issue; do let me know if you run into more problems.

[vimeitor]

I'm afraid I can still reproduce the issue with the same steps as before. Do not worry about it; there's no rush into fixing this (at least from my side). I might take a look at it in a few days if I get some time, although I have very little experience with vimscript besides the small functions that I have here and there on my vimrc.