Warning: commit message does not conform to UTF-8

[rendered]

I get this warning message alot:

Warning: commit message does not conform to UTF-8. You may want to amend it after fixing the message, or set the config variable il8n.commitencoding to the encoding your project uses.

Anything to be worried about?

[wimh]

I have seen it too, I do need to look at it, but so far I did not see anything fail because of this.

[hbons]

@wimh shall we set il8n.commitencoding to utf-8 to make this go away? The commit messages aren't important anyway.

[rendered]

Ok was just worried it might affect Sparkleshare.

[wimh]

@hbons I don't know exactly what il8n.commitencoding does, though I can guess it.

Note that the windows OS uses utf-16 for unicode, where linux uses utf-8. If I am correct, the commit message includes a non-ascii qoute (similar to `), so that may be the reason for the warning. So possibly it would be better for git to use utf-16 too. But I can't change that without testing if nothing breaks.....

There are more unicode issues which need attention. Because this is just a warning, I considered it more important to look at other issues first. If I leave this open, I won't forget....

[kyak]

Speaking about i18n issues, currently it doesn't work at all between a Linux machine (ru_RU.UTF-8) and a Windows machine (cp866 in cmd and cp1251 as system locale).

I.e. file names and commit messages where cyrillic is used are displayed as garbage in another OS, no matter how hard i play with i18n.commitencoding and i18n.logoutputencoding

[kyak]

Update: using this Git from here: http://code.google.com/p/utf8-git-on-windows/ solves the problem from git side (now file names are all right). However, there is still garbage at SparkeShare's "latest events" window (regarding the file names).

[wimh]

@kyak, I can't find sourcecode from that version. Just a binary download. Do you have any information about the differences between those git versions?

[kyak]

@wimh, yeah, seems strange that they don't provide source code for each binary. There is a source code for one of the binaries here: http://utf8-git-on-windows.googlecode.com/files/Git-1.7.3.2-UTF8-Source-20110110.7z I requested the author to provide the source code (for the latest binary), just in case.

BTW, the commit comment itself (containing the file name and coming from SparkleShare) is garbled, which leads to garbled text in "Latest events" window.

[kyak]

It seems that guys on msysgit side are taking the encoding problem seriously. "The issue 80" is well known around those who have such a problem: http://code.google.com/p/msysgit/issues/detail?id=80

The last message says about the ongoing work in https://groups.google.com/group/msysgit/browse_thread/thread/40112decdc564117. I currently use the v13 of unicode patch series for msysgit. It solves the problem with file names (whereas they are UTF-8 on my Linux machine and CP1251 or whatever on Windows machine) - at least i can confirm that cyrillic filenames display correctly on both machines.

The problem, however, remains in git commit message itself. The file name is displayed as garbade in commit message (both on Windows and Linux machines). I think this message is coming from SparkleShare.

I'm eager to help by testing this issue or providing more details. Thanks!

P.S. One more thing - the file name is displayed as garbage in SparkleShare's "Show Recent Events" interface, too.

P.P.S. More info: the same filenames in "Recent Events" display fine in Linux.

[snauth]

It seems that the issue is solved indeed with the latest msysgit preview http://code.google.com/p/msysgit/downloads/detail?name=Git-1.7.10-preview20120409.exe&can=2&q=