How to handle line endings in Git

[fandersen]

Hi everyone,

I am using WinCC OA 3.20 P007 in Windows.

I want to version my WinCC OA project in Git.

The default installtion of Git suggests to convert line endings CRLF to LF when committing and converting LF to CRLF when checking out.

Is this behavior OK with WinCC OA or should I better select "Checkout as-is, commit as-is"?

There are no linux developers involved.

BR
Florian

[hpuchegger]

We use the following settings in ".gitattributes":

Code: Select all

* text=auto
# panel files should not be CRLF changed since LangText format contains number of BYTES to read
*.pnl -text

# config binary
*.tar binary
*.dpl text=auto

auto => CRLF 2 LF
text => no conversion
binary => no conversion

Br, Herbert

[fandersen]

Hi Herbert,

thank you for your valuable input!

I will use this as a starting point and figure out what file types we have and how to them.


BR
Florian

[fandersen]

Hi Herbert,

in your example you treat *.dpl files as regular text.

But, as there are also *number of bytes to read*, couldn't it happen there too, that turning LF to CRLF breaks something?

BR
Florian

[gschijndel]

The dpl file is a tab separated values file. There should be no number of bytes to read in it.

But when you are busy with the .gitattributes you should also consider to enable git-lfs for binary files. It makes cloning faster due to a smaller repo size.
In case linguist is used to determine the language (by gitlab/github for example) you can tell it to use another language for the scripts and libraries.

Code: Select all

# Use git-lfs for these file types
*.dll	filter=lfs diff=lfs merge=lfs -crlf
*.exe	filter=lfs diff=lfs merge=lfs -crlf
*.pdf	filter=lfs diff=lfs merge=lfs -crlf
*.so	filter=lfs diff=lfs merge=lfs -crlf

# Prevent that as programming language visual basic is shown
*.ctl linguist-language=C++

[fandersen]

Hi gschijndel,

I have checked the dpl format. And you are right. In the format "Regular export", there are no "number of bytes to read" in it. By default, the ASCII file is already created with CRLF eol by the ASCII manager.

It is different for Format 2 (Multi language export) of the ASCII export. We use it for some special cases. There you have "number of bytes to read". But I could not get RAIMA to store translations with line breaks. So I guess I don't need to worry about corruption.

About, git-lfs: Yeah we heard about it. Git stores the binary files on the Git LFS server and puts a pointer to the file inside the repo instead. We definitively check that out, thanks!

BR
Florian

[leoknipp]

The ASCII format 2 was used in the past as it was still possible to have language dependent names for DP types and datapoints.
As it is hard to read and not easy to edit we recommend not to use this format.

Best Regards
Leopold Knipp
Senior Support Specialist