git@vger.kernel.org mailing list mirror (one of many)
 help / color / mirror / code / Atom feed
From: "Torsten Bögershausen" <tboegi@web.de>
To: Mr&Mrs D <the.ubik@gmail.com>
Cc: git <git@vger.kernel.org>
Subject: Re: Bug in lineendings handling that prevents resetting checking out, rebasing etc
Date: Sat, 15 Dec 2018 05:43:45 +0100	[thread overview]
Message-ID: <20181215044345.GA19192@tor.lan> (raw)
In-Reply-To: <CABRG_PG+D73XPDy3RV1-5yJvSuKDV9ymjT6pp5Jt5DoVsLZT8A@mail.gmail.com>

On Fri, Dec 14, 2018 at 04:51:54PM -0500, Mr&Mrs D wrote:
> Thanks for looking to it - git attributes was added in
> 4b0863a8b834c5804fd3a568ed56ff85b27acdeb
> 
> The file in question was added in 17ca75f0a8c25f321f2b63bc3b9c065ff91adc23
> 
> So you mean to say that because a gitattributes was added after the
> fact this resulted in an illegal state?
> 
> But _shouldn't git reset --hard work anyway?_ That's the buggy part.

Yes and no.
If I look at the dev branch:
commit 02ae6f264f340137b8b41ba6953e2a4f962c222e (HEAD, origin/dev, origin/HEAD, dev)

Now we can ask Git, why a file is modified:

git ls-files --eol | grep "Mopy/Docs/Bash eadme Template.html"
i/crlf  w/crlf  attr/text               Mopy/Docs/Bash Readme Template.html

Now we have 2 conflicting "messages" to Git:
a) "Mopy/Docs/Bash Readme Template.html" has the attribute "text"
b) "Mopy/Docs/Bash Readme Template.html" has been commited with CRLF.

Git itself can not resolve this conflict.
Either you normalize the repo (in this case only 1 file), other commits have 4 files
that needs to be normalized.
Or you change the attribute into "text=auto".

That decision is up to the user.

> 
> As for fixing it - not sure what is the best course of action here.
> probably issuing `git add --renormalize .` and committing that to the
> stable (dev) branch will fix this for future checkouts/rebases but
> IIUC won't do nothing for older commits - so checking out a commit
> before the fix one, ghit will see this file as changed and then
> completely refuse to go back to another branch
> 
> This seems a bug - as illegal as the state of the file is, shouldn't
> git reset always work?
> 
> Thanks! (will be out for a bit but really looking forward to your replies)
> On Fri, Dec 14, 2018 at 4:32 PM Torsten Bögershausen <tboegi@web.de> wrote:
> >
> > On Fri, Dec 14, 2018 at 04:04:15PM -0500, Mr&Mrs D wrote:
> > > Hi all,
> > >
> > > I maintain a python project you can clone from:
> > >
> > > git@github.com:wrye-bash/wrye-bash.git
> > >
> > > For reasons unknown git sees a particular file as changed
> > > (Mopy/Docs/Bash Readme Template.html, sometimes others too). This file
> > > was probably committed to the svn repository this git repo was created
> > > from with CRLF line endings. When we moved to git we added a
> > > gitattributes file (
> > > https://github.com/wrye-bash/wrye-bash/blob/dev/.gitattributes ) and
> > > this file was edited to explicitly state htms are text - all to no
> > > avail. From time to time - on windows - as in when checking out an old
> > > commit - git would see that file as changed. The only workaround that
> > > worked for me was
> > >
> > >     git rm -r . --cached -q && git reset --hard
> > >
> > > For more details and discussion see this SO question I posted almost
> > > five years ago:
> > >
> > > https://stackoverflow.com/questions/21122094/git-line-endings-cant-stash-reset-and-now-cant-rebase-over-spurious-line-en
> > >
> > > I used to work in windows and the bug was tolerable as there was that
> > > workaround. Now I moved to mac and no workaround works anymore - we
> > > have a special page on our wiki  with workarounds for this one btw:
> > >
> > > https://github.com/wrye-bash/wrye-bash/wiki/%5Bgit%5D-Issues-with-line-endings-preventing-checking,-merge,-etc
> > >
> > > Well after 5 years and countless hours trying to solve this I reach
> > > out to you guys and girls - _this is a full-time bug in git line
> > > endings handling_. When someone issues a git reset --hard this should
> > > work no matter what - well it does not. So this bug may be really a
> > > can of worms.
> > >
> > > Please someone clone this repo on linux or mac - probably just cloning
> > > will have the files appear as changed (by the way hitting refresh on
> > > git gui I have different sets of files appear as changed). If not then
> > >
> > > git checkout utumno-wip
> > Thet commit is -excuse me if that sounds too harsh- is messed up.
> > git status says
> > modified:   Mopy/Docs/Bash Readme Template.html
> >
> > And if I dig into the EOL stuff, I run
> > git ls-files --eol | grep  Readme | less
> >
> > And find a contradiction here:
> > i/crlf  w/crlf  attr/text               Mopy/Docs/Bash Readme Template.html
> >
> > The attributes say "text" and the file has CRLF "in the repo",
> > (techically speaking in the index) and that is an "illegal" condition
> > in the repo, and not a bug in Git.
> > I didn't try the rebase as such, sice the first problem needs
> > to be fixed, before we try to move on.
> >
> > So, the old commits are problematic/illegal and they are as they are.
> > Such a commit can not be fixed, whenever somebody checks it out,
> > there will be a problem (or two, or none, depending on the timing,
> > the file system...)
> >
> > We can not fix commits like b1acc012878c9fdd8b4ad610ce7eae0dcbcbcab0.
> > We can make new commits, and fix them.
> >
> > We can fix one branch, and other branches, and merge them together.
> > But rebase seems to be problamatic, at least to me.
> > What exactly do you want to do?
> >
> > Can we agree to do a merge of 2 branches?
> > Then I can possibly help you out.
> >
> >
> >
> >
> >
> > > git rebase -i dev
> > >
> > > and then select a commit to edit should be enough to trigger this bug
> > >
> > > Needless to say I am  well aware of things like `git add --renormalize
> > > .` - but renormalizing is not the issue. The issue is that _files show
> > > as changed and even a git reset --hard won't convince git that
> > > nothing's changed_.
> > >
> > > $ git reset --hard
> > > HEAD is now at e5c16790 Wip proper handling of ini tweaks encoding - TODOs:
> > > $ git status
> > > interactive rebase in progress; onto 02ae6f26
> > > Last commands done (4 commands done):
> > >    pick 3a39a0c0 Monkey patch for undecodable inis:
> > >    pick e5c16790 Wip proper handling of ini tweaks encoding - TODOs:
> > >   (see more in file .git/rebase-merge/done)
> > > Next commands to do (19 remaining commands):
> > >    edit a3a7b237 Amend last commit and linefixes:  ΕΕΕΕ
> > >    edit 432fd314 fFF handle empty or malformed inis
> > >   (use "git rebase --edit-todo" to view and edit)
> > > You are currently editing a commit while rebasing branch 'utumno-wip'
> > > on '02ae6f26'.
> > >   (use "git commit --amend" to amend the current commit)
> > >   (use "git rebase --continue" once you are satisfied with your changes)
> > >
> > > Changes not staged for commit:
> > >   (use "git add <file>..." to update what will be committed)
> > >   (use "git checkout -- <file>..." to discard changes in working directory)
> > >
> > > modified:   Mopy/Docs/Bash Readme Template.html
> > >
> > > Untracked files:
> > >   (use "git add <file>..." to include in what will be committed)
> > >
> > > .DS_Store
> > > .idea.7z
> > >
> > > no changes added to commit (use "git add" and/or "git commit -a")
> > > $
> > >
> > > I really hope someone here can debug this
> > > Thanks!

  reply	other threads:[~2018-12-15  4:43 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-12-14 21:04 Bug in lineendings handling that prevents resetting checking out, rebasing etc Mr&Mrs D
2018-12-14 21:21 ` John Passaro
2018-12-14 21:32   ` Mr&Mrs D
2018-12-14 21:32 ` Torsten Bögershausen
2018-12-14 21:51   ` Mr&Mrs D
2018-12-15  4:43     ` Torsten Bögershausen [this message]
2018-12-15 20:58       ` Mr&Mrs D

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: http://vger.kernel.org/majordomo-info.html

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20181215044345.GA19192@tor.lan \
    --to=tboegi@web.de \
    --cc=git@vger.kernel.org \
    --cc=the.ubik@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public inbox

	https://80x24.org/mirrors/git.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).