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. > 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 > <https://groups.google.com/d/msgid/sage-devel/0e364799-33a8-4623-8faa-c0d8a17f5ac4n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- 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.
