vimagit can be very slow

[mg979]

Short description of the problem

vimagit buffer update can take ages. It usually happens when there are a lot of changes. My guess is that it has to do with folding, because if I abort with <C-c> while the process is stuck, close the buffer and reopen vimagit, hunks are staged/unstaged correctly. Maybe vimagit updates the buffer while folding is active, bit by bit? In this case a possible solution would be to disable folding while updating buffer, or similar.

Steps to reproduce

The more text, the slower it is. It's not dependent on the number of hunks, but on how big hunks are.

Environment

Linux Debian stable.

• vim or nvim version: vim 8.0+ (sometimes 8.1)

• git version: 2.11

• current vimagit version or (or SHA1): 85c25ff

• terminal or gvim: terminal

[jreybert]

Thanks for your report, this is a lead I never explored.
I'll dig around it and update this ticket if I find anything relevant.

[mg979]

Maybe it has also to do with certain filetypes, because usually it's fast. It behaved worst with dos files with very long lines and lots of tabs (a sort of table).

[jreybert]

As you mention very long lines, it reminds me that vim by itself can be very slow with long lines, because of syntax coloring.

[jreybert]

By any chance, would you have a public repository which makes vimagit slow?

Could you also test with this following setting in your vimrc:
let g:magit_default_fold_level = 2

[mg979]

You could try with this commit, make a branch out of it, then git reset --hard HEAD^ and squash the original branch without committing, so that you can have the unstaged commit in vimagit.

The README.md file takes a lot to be staged/unstaged, but hit <c-c>, quit vimagit, and enter again, you'll see that the operation is done, and vimagit loads fast as usual.

Edit: let g:magit_default_fold_level = 2 doesn't help.

[jreybert]

Excellent, thanks a lot for your input!

[jreybert]

Hourray! It has been fixed!

Thanks a lot for all your input, it wouldn't have been possible without that.

Could you try the branch next?

[mg979]

Thanks, I'll be able to test it in a few days because I have only Windows in this machine and gvim there doesn't seem to have this problem (even if it's slow on its own because command line stuff seems to much slower on Windows anyway), but I'm confident it will work :)

[mg979]

Ok I tested it and it seems perfect to me, much more responsive, thanks!