Re: Lost files after git stash && git stash pop

[Phillip Wood <phillip.wood123@gmail.com>]

On 22/07/2023 22:44, Torsten Bögershausen wrote:
> On Fri, Jul 21, 2023 at 07:31:53PM +0200, Till Friebe wrote:
>> Thank you for filling out a Git bug report!
>> Please answer the following questions to help us understand your issue.
>>
>> What did you do before the bug happened? (Steps to reproduce your issue)
>> ```
>> git init
>> mkdir README
>> touch README/README
>> git add .
>> git commit -m "Init project"
>> echo "Test" > README/README
>> mv README/README README2
>> rmdir README
>> mv README2 README
>> git stash
>> git stash pop
>> ```
>>
>> What did you expect to happen? (Expected behavior)
>> I expected that after the `git stash pop` the README file would be back.
>>
>> What happened instead? (Actual behavior)
>> This README with "Test" file was deleted and I lost 5 hours of work.
> 
> That is always sad to hear, when work is lost.

Indeed it is. Thanks Till for providing an easy reproducer.

> However, I personally wonder if this is a bug or not.

I think whenever git overwrites an untracked file without the user 
passing some option indicating that they want to do so it is a bug. For 
example "git checkout" refuses to overwrite untracked files by default. 
Sadly this seems to be a known bug in do_push_stash() where we are using 
"git reset --hard" to remove the stashed changes from the working copy. 
This was documented in 94b7f1563a (Comment important codepaths regarding 
nuking untracked files/dirs, 2021-09-27). The stash implementation does 
a lot of necessary forking of subprocesses, in this case I think it 
would be better to call unpack_trees() directly with 
UNPACK_RESET_PROTECT_UNTRACKED.

Best Wishes

Phillip

> First, Git is told to track a file called README/README
> Then the file is removed, without telling Git.
> And a new, unkown file appers on disk (which collides with the name
> of the directory)
> 
> Using this sequence could have told Git, what is going on:
> git mv README/README README2
> rmdir README
> git mv README2 README
> 
> (a temporary branch may be checked out, with the option
>   to merge-squash the final result)
> 
> 
> An other alternative could be to tell `git stash` to care
> about untracked file(s):
> 
> git stash -u
> git stash pop
> 
> Which will refuse to apply the stash.
> 
> A third alternative could be to keep the file inside an
> editor, to have the content still available.
> 
> However, it would/could be nice, if files are not simply deleted,
> but saved into a "lost+found" folder, or a wastebasket kind of thing.
> 
> But which files ?
> Those that are untracked ?
> They may be important (local config files, passwords, help scripts, ...)
> or not (.o files from a C compiler).
> 
> In some older discussions they had been named "precious" files.
> But, as far as I remember, there was no easy solution.
> In that sense I don't have a better answer.
> Others may have.
> 
> Thanks for reporting, it make me read [1] and come to the conclusion
> that it is sometimes safer to checkout out a temporary branch, commit
> everything and clean up later, rather than relying too much on
> `git stash`
> 
> 
> <https://stackoverflow.com/questions/835501/how-do-you-stash-an-untracked-file>
>>
>> What's different between what you expected and what actually happened?
>> The file doesn't exist anymore and I can't recover it.
>>
>> Anything else you want to add:
>> This is just a reproducible example.
>>