Re: [PATCH v4] sparse-checkout.txt: new document with sparse-checkout directions

[Elijah Newren <newren@gmail.com>]

On Sat, Jan 14, 2023 at 2:19 AM ZheNing Hu <adlternative@gmail.com> wrote:
>
> Elijah Newren <newren@gmail.com> 于2022年11月16日周三 13:49写道：
>
> > [...]
> > > > +    The fact that files can move between the 'tracked' and 'untracked'
> > > > +    categories means some commands will have to treat untracked files
> > > > +    differently.  But if we have to treat untracked files differently,
> > > > +    then additional commands may also need changes:
> > > > +
> > > > +      * status
> > > > +      * clean
> > > > +
> > >
> > > I'm a bit worried about git status, because it's used in many shells
> > > (e.g. zsh) i
> > > in the git prompt function. Its default behavior is restricted, otherwise users
> > > may get blocked when they use zsh to cd to that directory. I don't know how
> > > to reproduce this problem (since the scenario is built on checkout to a local
> > > unborn branch).
> >
> > Could you elaborate?  I'm not sure if you are talking about an
> > existing problem that you are worried about being exacerbated, or a
> > hypothetical problem that could occur with changes.  Further, your
> > wording is so vague about the problem, that I have no idea what its
> > nature is or whether any changes to status would even possibly have
> > any bearing on it.  But the suggested changes to git status are
> > simply:
> >
>
> I find this special case, it will fetch some blobs when "git status".
> First, we init a git repository, then set sparse specification to "*.js" with
> no-cone mode, then use blob:none filter to fetch all commits and trees,
> and finally checkout to the default branch.
>
> #!/bin/sh
>
> rm -rf sparse-checkout-example
> git init sparse-checkout-example
> git -C sparse-checkout-example remote add origin
> git@github.com:derrickstolee/sparse-checkout-example.git
> git -C sparse-checkout-example sparse-checkout set --no-cone *.js
> git -C sparse-checkout-example fetch origin --filter=blob:none main
> git -C sparse-checkout-example branch --track main origin/main
> git -C sparse-checkout-example checkout main
>
> Then let's do a git status, which some zsh git plugin
> will do when user "cd" the git repository.
>
> # git -C sparse-checkout-exmaple status
>
> remote: Enumerating objects: 1, done.
> remote: Total 1 (delta 0), reused 0 (delta 0), pack-reused 1
> Receiving objects: 100% (1/1), 416 bytes | 416.00 KiB/s, done.
> remote: Enumerating objects: 1, done.
> remote: Total 1 (delta 0), reused 0 (delta 0), pack-reused 1
> Receiving objects: 100% (1/1), 160 bytes | 160.00 KiB/s, done.
> remote: Enumerating objects: 1, done.
> remote: Total 1 (delta 0), reused 0 (delta 0), pack-reused 1
> Receiving objects: 100% (1/1), 2.01 KiB | 2.01 MiB/s, done.
> remote: Enumerating objects: 1, done.
> remote: Total 1 (delta 0), reused 0 (delta 0), pack-reused 1
> Receiving objects: 100% (1/1), 120 bytes | 120.00 KiB/s, done.
> On branch main
> Your branch is up to date with 'origin/main'.
>
> You are in a sparse checkout with 8% of tracked files present.
>
>
> It took 17.00 seconds to enumerate untracked files. 'status -uno'
> may speed it up, but you have to be careful not to forget to add
> new files yourself (see 'git help status').
> nothing to commit, working tree clean
>
> Yeah, here it fetches four blobs, and takes 17s!!!
>
> So what blobs are we fetching?
>
> GIT_TRACE_PACKET=1 git -C sparse-checkout-example status
> ...
> 18:02:32.989231 pkt-line.c:80           packet:        fetch> want
> dff85a65c0ef4b50a4c01bdd4a247b974bc45f90
> ...
> 18:02:37.059203 pkt-line.c:80           packet:        fetch> want
> f07ead02d13f62414589b1f1b891bb6a764ec91f
> ...
> 18:02:40.868899 pkt-line.c:80           packet:        fetch> want
> 3c4efe206bd0e7230ad0ae8396a3c883c8207906
> ...
> 18:02:44.961809 pkt-line.c:80           packet:        fetch> want
> 6590681af7e177dc71fe08648c4bbf4223b82866
>
>
> Then let's we look what's the blob:
> git log --find-object=dff85a65c0ef4b50a4c01bdd4a247b974bc45f90 --stat
> commit 8ec229339caad56eb849c67361a9699004564177
> Author: Derrick Stolee <dstolee@microsoft.com>
> Date:   Mon Dec 30 13:30:27 2019 -0500
>
>     Add twbs/bootstrap
>
>  web/browser/.gitignore | 47 +++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 47 insertions(+)
>
> git log --find-object=f07ead02d13f62414589b1f1b891bb6a764ec91f --stat
> commit 9c5a31030de62355410a322923e33e90a00032f6
> Author: Derrick Stolee <dstolee@microsoft.com>
> Date:   Mon Dec 30 13:31:06 2019 -0500
>
>     Add artsy/artsy.github.io
>
>  web/editor/.gitignore | 13 +++++++++++++
>  1 file changed, 13 insertions(+)
>
> Yeah, it seems that git status fetch these .gitigore files.
> So what's the wrong here?

Nothing; this looks exactly as I'd expect.

`git status` is supposed to check whether untracked files are ignored
or not, and it needs .gitignore files to do so.  If an entire
directory is missing, then it can avoid loading a .gitignore under
that directory, but you set up your patterns such that no directory is
likely to be excluded.

Maybe there are other special cases that one could attempt to handle
(e.g. first check if there are any untracked files in a directory and
only then check for which are ignored, but that'd probably take some
big refactoring of some hairy dir.c code to accomplish something like
that, and you'd have to be really careful to not regress the
performance of non-sparse cases).  Personally, I don't think it's
worth it.  If you really don't like it, I'd suggest choosing any one
of the following ways to avoid it:

  * Include the .gitignore files in your sparse-checkout; might as
well since you're going to get them anyway.
  * Set status.showUntrackedFiles to `no` so that .gitignore files are
irrelevant
  * Use cone mode instead of non-cone mode (and yes, restructure your
repo if needed)
  * Remove the .gitignore files and commit their deletion.  Just do
without them, somehow.