Re: [PATCH v1 1/1] git stash needing mkdir deletes untracked file

From: Phillip Wood <phillip.wood123@gmail.com>
Date: Aug 15, 2023, 09:15 UTC
Message-ID: <9f76de24-d337-ed41-fb81-888dba0b1656@gmail.com>
In-Reply-To: <20230809184751.ffwolkvjwoptnmen@tb-raspi4>

Hi Torsten

Sorry for the slow reply

On 09/08/2023 19:47, Torsten Bögershausen wrote:

Show 15 quoted lines

> On Wed, Aug 09, 2023 at 02:15:28PM +0100, Phillip Wood wrote:
>> Hi Torsten
>>
>> Thanks for working on this. I've cc'd Junio for his unpack_trees()
>> knowledge.
> 
> Thanks Eric for the review.
> 
> Hej Phillip,
> I have been playing around with the whole thing some time.
> At the end I had a version, which did fiddle the information
> that we are doing a `git stash` (and not any other operation)
> into entry.c, and all test cases passed.
> So in principle I can dig out all changes, polish them
> and send them out, after doing cleanups of course.

I don't think we should be treating "git stash" as a special case here - commands like "git checkout" should not be removing untracked files unprompted either.

Show 6 quoted lines

> (And that could take a couple of days, or weeks ;-)
> 
> My main question is still open:
> Is it a good idea, to create a "helper file" ?
> The naming can be discussed, we may stick the date/time
> into the filename to make it really unique, or so.

I think stopping and telling the user that the file would be overwritten as we do in other cases would be better.

> Reading the different reports and including own experience,
> I still think that a directory called ".deleted-by-user"
> or ".wastebin" or something in that style is a good idea.

I can see an argument for being able to opt-in to that for "git restore" and "git reset --hard" but that is a different problem to the one here.

Best Wishes

Phillip