On Thu, Feb 11, 2021 at 10:43 AM John Cremona <john.crem...@gmail.com> wrote:
>
>
>
> On Thu, 11 Feb 2021, 10:00 Frédéric Chapoton, <fchapot...@gmail.com> wrote:
>>
>> I have made https://trac.sagemath.org/ticket/31379 to get rid of the CRLF in
>> these 3 files.
I guess one can ensure that CRLF get automatically converted into LF
on git push to trac,
for all files which are not specifically declared as binary in .gitattributes
I believe that the trouble came from pulling from a github repo, as
github might have its own default ideas
about dealing with CRLF...
>
>
> Thanks!
>
> Can we do something to stop this happening again?
>
> John
>
>
>>
>> Le jeudi 11 février 2021 à 10:43:27 UTC+1, Frédéric Chapoton a écrit :
>>>
>>> I have been hit by this kind of annoying issue some time ago. No idea what
>>> is happening.
>>> There are only 3 offending files containing CRLF:
>>>
>>> sage$ git grep -Il $'\r' src/
>>> src/sage/combinat/crystals/star_crystal.py
>>> src/sage/misc/element_with_label.py
>>> src/sage/rings/invariants/__init__.py
>>>
>>> We should probably change them to use the correct line break
>>> Le jeudi 11 février 2021 à 10:21:26 UTC+1, john.c...@gmail.com a écrit :
>>>>
>>>> On Wed, 10 Feb 2021 at 18:07, Dima Pasechnik <dim...@gmail.com> wrote:
>>>> >
>>>> > On Wed, Feb 10, 2021 at 5:49 PM John Cremona <john.c...@gmail.com> wrote:
>>>> > >
>>>> > > On Wed, 10 Feb 2021 at 16:56, Dima Pasechnik <dim...@gmail.com> wrote:
>>>> > > >
>>>> > > > On Wed, Feb 10, 2021 at 2:58 PM John Cremona <john.c...@gmail.com>
>>>> > > > wrote:
>>>> > > > >
>>>> > > > > Has anyone else been seeing the following problem, which has been
>>>> > > > > plaguing me for a week or two. Here's a simple example. All
>>>> > > > > computers mentioned here are running ubuntu. On a machine I had not
>>>> > > > > used for a while I had a sage build of the develop branch at an old
>>>> > > > > version (pre 9.0). There were no modified files (git status showed
>>>> > > > > nothing). Then I did "git pull trac develop", after which one file
>>>> > > > > is
>>>> > > > > marked as having changed:
>>>> > > > >
>>>> > > > > modified: src/sage/misc/element_with_label.py
>>>> > > > >
>>>> > > > > It is always this file, on several machines where I have gone
>>>> > > > > through
>>>> > > > > similar steps. file shows this:
>>>> > > > > src/sage/misc/element_with_label.py: Python script, ASCII text
>>>> > > > > executable, with CRLF line terminators
>>>> > > >
>>>> > > > does
>>>> > > >
>>>> > > > git config core.autolf
>>>> > > >
>>>> > > > show 'true'?
>>>> > >
>>>> > > No.
>>>> > >
>>>> > > >
>>>> > > > Set it to false, IMHO this should fix this issue.
>>>> > >
>>>> > > It is not set at all. I saw that option after googling for help, but
>>>> > > the issue it exists to solve is not one which has ever hit me (in sage
>>>> > > anyway, of course I know about using dos2unix sometimes when a windows
>>>> > > users sends a file).
>>>> >
>>>> > did you try setting this option to false, and see if it helps?
>>>>
>>>> Not yet systematically. After setting it to false and doing nothing
>>>> else, git status still shows
>>>> src/sage/misc/element_with_label.py as modified, even after 'git
>>>> checkout --'. Trying to get back to sanity a different way, 'git
>>>> stash' now outputs
>>>>
>>>>
>>>> warning: CRLF will be replaced by LF in
>>>> src/sage/misc/element_with_label.py.
>>>> The file will have its original line endings in your working directory.
>>>> warning: CRLF will be replaced by LF in
>>>> src/sage/misc/element_with_label.py.
>>>> The file will have its original line endings in your working directory.
>>>> Saved working directory and index state WIP on develop: 8453ffb
>>>> Updated SageMath version to 9.3.beta7
>>>> HEAD is now at 8453ffb Updated SageMath version to 9.3.beta7
>>>>
>>>> but git status still shows that file as modified. Now I cannot just
>>>> checkout a different existing branch (since there are apparently
>>>> modified files), but I can checkout a new branch (g = alias for git)
>>>>
>>>> $ g co -b dud
>>>> M src/sage/misc/element_with_label.py
>>>> Switched to a new branch 'dud'
>>>>
>>>> the delete the 'develop' branch (which was the same as upstream anyway):
>>>>
>>>> $ g branch -d develop
>>>> Deleted branch develop (was 8453ffb).
>>>>
>>>> and now recreate the develop branch
>>>>
>>>> $ g remote update trac
>>>> (...)
>>>> $ g co -b develop trac/develop
>>>> M src/sage/misc/element_with_label.py
>>>> Branch develop set up to track remote branch develop from trac.
>>>> Switched to a new branch 'develop'
>>>>
>>>> and we are back to where we started. All of the above was with
>>>> core.autolf set to false.
>>>>
>>>> On the same machine, in a completely new clone made and built
>>>> yesterday, git status shows
>>>>
>>>> $ g st
>>>> On branch develop
>>>> Your branch is up-to-date with 'origin/develop'.
>>>> 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: src/sage/rings/invariants/__init__.py
>>>>
>>>> no changes added to commit (use "git add" and/or "git commit -a")
>>>>
>>>> So the same problem with a new file -- which only has 2 bytes in it!
>>>> AFter dos2unix-ing it (so it now has 1 byte) it is still showing up in
>>>> git status. AFter git checkout -- it goes back to having 2 bytes and
>>>> still shows up.
>>>>
>>>> It seems to be impossible to do any sage development here. I am now
>>>> trying yet another git clone -- so the global config of core.autolf is
>>>> in force as the clone is made -- but then after the clone from
>>>> git://github.com/sagemath/sage.git that same file is shown as
>>>> modified. For some reason I have been making the cones from github
>>>> and then adding trac as a remote. As a variant I made a fresh clone
>>>> from trac -- but see exactly the same.
>>>>
>>>> A similar problem was reported on stackexchange *over 9 years ago* by
>>>> a Mac users whose collaborators were all ubuntu users. There, it is
>>>> suggest to look at the .gitattributes file, and *in this fresh clone*
>>>> that file contains
>>>>
>>>> # Force LF normalization
>>>> * text=auto eol=lf
>>>> # except for Windows batch files
>>>> *.{cmd,[cC][mM][dD]} text eol=crlf
>>>> *.{bat,[bB][aA][tT]} text eol=crlf
>>>> *.diff_bin binary
>>>>
>>>> This is commit 5328570 which is 9.2, master branch, but the develop
>>>> branch has the same.
>>>>
>>>> To me it seems (from reading all that stackexchange page) that this is
>>>> not just a problem caused by something in my local configuration, but
>>>> something which has happened to the repository itself, which needs to
>>>> be fixed. But I am no expert.
>>>>
>>>> Just to add to the fun, on a second machine with the same ubuntu
>>>> version as the one above (16.04.7 LTS) and same git version 2.7.4, a
>>>> clone I made yesterday shows no problems at all, while and older clone
>>>> on which I had done development for a long time has some CRLF file
>>>> issues. The only differences in the global .gitconfig between the two
>>>> machines is the core.autolf which I just added to the first.
>>>>
>>>> I do have some clones without these problems so can keep on working,
>>>> but I also have branches I wanted to do more work on in clones which
>>>> are unusable.
>>>>
>>>> John
>>>>
>>>>
>>>> >
>>>> > >
>>>> > > So, I did not change anything (as far as I can tell -- except get
>>>> > > older) but my git state changed. These files stay changed even after
>>>> > > trying to reset them (though *I* did not change them) using git
>>>> > > checkout --.
>>>> >
>>>> > It's a new version of git that got you, I suppose.
>>>>
>>>> Yes, that is likely, since I did do package updates across the board
>>>> before this started to happen. I am seeing it on machines running:
>>>> ubuntu 16.04.7 LTS and git 2.7.4
>>>> ubuntu 18.04.5 LTS and git 2.17.1
>>>> ubuntu 20.04.2 LTS and git 2.25.1
>>>>
>>>> In no cases was the whole OS updated, just packages so possibly git,
>>>> but I don't know what git version any had before this.
>>>> >
>>>> > We've had some "fun" with this lately, as there are Sage spkgs which
>>>> > need patches for files
>>>> > with CRLF endings, and so these patches had to be dealt with somehow.
>>>> > It was not pleasant.
>>>> > See https://trac.sagemath.org/ticket/30403
>>>> >
>>>> >
>>>> > >
>>>> > > John
>>>> > >
>>>> > > >
>>>> > > >
>>>> > > > >
>>>> > > > > git diff shows this:
>>>> > > > >
>>>> > > > > $ git diff -w src/sage/misc/element_with_label.py
>>>> > > > > warning: CRLF will be replaced by LF in
>>>> > > > > src/sage/misc/element_with_label.py.
>>>> > > > > The file will have its original line endings in your working
>>>> > > > > directory.
>>>> > > > >
>>>> > > > > I cannot get this to go away. git stash does nothing, and I cannot
>>>> > > > > change branches since git thinks there is an uncommitted change. I
>>>> > > > > have been reduced to deleting the entire build and starting again
>>>> > > > > from
>>>> > > > > scratch.
>>>> > > > >
>>>> > > > > No, I have not knowingly changed any configuration on these
>>>> > > > > computers,
>>>> > > > > though I have done some normal package upgrades using apt.
>>>> > > > >
>>>> > > > > I just went through the same steps in another directory on the same
>>>> > > > > machine where I had an old build (8.7) and after "git pull trac
>>>> > > > > develop" there are now 5 modified files:
>>>> > > > >
>>>> > > > > modified:
>>>> > > > > build/pkgs/sage_sws2rst/src/test/Adding_Pictures_and_screenshots.sws
>>>> > > > > modified: src/doc/en/reference/combinat/media/k-rim.JPG
>>>> > > > > modified: src/sage/ext_data/doctest/rich_output/example.png
>>>> > > > > modified: src/sage/ext_data/notebook-ipython/logo-64x64.png
>>>> > > > > modified: src/sage/misc/element_with_label.py
>>>> > > > >
>>>> > > > > -- the same one as before and 3 others which are not even normal
>>>> > > > > text
>>>> > > > > files. I don't want to trash all these since I am trying to
>>>> > > > > resurrect
>>>> > > > > and complete work done on some old trac tickets.
>>>> > > > >
>>>> > > > > John
>>>> > > > >
>>>> > > > > --
>>>> > > > > You received this message because you are subscribed to the Google
>>>> > > > > Groups "sage-devel" group.
>>>> > > > > To unsubscribe from this group and stop receiving emails from it,
>>>> > > > > send an email to sage-devel+...@googlegroups.com.
>>>> > > > > To view this discussion on the web visit
>>>> > > > > https://groups.google.com/d/msgid/sage-devel/CAD0p0K7KrVUvvkmqt6Nwhe1Tt9ef_w3ZWAqU5GK6Y6%3Dysqq3Uw%40mail.gmail.com.
>>>> > > >
>>>> > > > --
>>>> > > > You received this message because you are subscribed to the Google
>>>> > > > Groups "sage-devel" group.
>>>> > > > To unsubscribe from this group and stop receiving emails from it,
>>>> > > > send an email to sage-devel+...@googlegroups.com.
>>>> > > > To view this discussion on the web visit
>>>> > > > https://groups.google.com/d/msgid/sage-devel/CAAWYfq10WrFPb-2wQfLVyhSSRRGdWr_eEQ7gyanQ9Pbgm6o6qQ%40mail.gmail.com.
>>>> > >
>>>> > > --
>>>> > > You received this message because you are subscribed to the Google
>>>> > > Groups "sage-devel" group.
>>>> > > To unsubscribe from this group and stop receiving emails from it, send
>>>> > > an email to sage-devel+...@googlegroups.com.
>>>> > > To view this discussion on the web visit
>>>> > > https://groups.google.com/d/msgid/sage-devel/CAD0p0K41axYDsedDj4-Dc7TkANM7T3CmVyMmS%3DDHhEXQQZqn4A%40mail.gmail.com.
>>>> >
>>>> > --
>>>> > You received this message because you are subscribed to the Google
>>>> > Groups "sage-devel" group.
>>>> > To unsubscribe from this group and stop receiving emails from it, send
>>>> > an email to sage-devel+...@googlegroups.com.
>>>> > To view this discussion on the web visit
>>>> > https://groups.google.com/d/msgid/sage-devel/CAAWYfq19RqXJ1oNgiTo3hX0yV06dXtMEenu-yuXjoM5S9XLhfg%40mail.gmail.com.
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "sage-devel" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to sage-devel+unsubscr...@googlegroups.com.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/sage-devel/0e364799-33a8-4623-8faa-c0d8a17f5ac4n%40googlegroups.com.
>
> --
> You received this message because you are subscribed to the Google Groups
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sage-devel+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sage-devel/CAD0p0K5YysFPDs8Z%3DWGg7%2B7Vyz0kP-OUgPFMMbWVgybC2e8ynA%40mail.gmail.com.
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/sage-devel/CAAWYfq0q%2Bwzn7D8UHwzF0V-nLmj-OHD%3DQuecD8E_-RVi41tuqg%40mail.gmail.com.
