hg cat command is executed during content indexing with filename passed as an argument and while Windows converts arguments to ANSI codepage, garbled filename gets to git process, eg:
> hg --config ui.verbose=false cat -r 547a7f4d4890c363836ebe7ae0bfb7e2dc421095 αφιλε.τχτ ?????.???: no such file in rev 547a7f4d4890
This leads to an error during indexing file content being printed to log and visible in Administration > Repository Status > Recent Errors section:
Problem indexing content of "αφιλε.τχτ" for rev "547a7f4d4890c363836ebe7ae0bfb7e2dc421095" from repo "unicode-file-names" due to class com.atlassian.fisheye.dvcs.handler.DvcsProcessException - Error while communicating with VCS: ???.?: no such file in rev 547a7f4d4890
and also an attempt to display file content (request for /browse/unicode-file-names/%CE%B1%CF%86%CE%B9%CE%BB%CE%B5.%CF%84%CF%87%CF%84?r=547a7f4d4890c363836ebe7ae0bfb7e2dc421095) cause an error to be displayed.
Described problem occurs if file.encoding is set to UTF-8 (it's a value that we recommend user to force during fisheye startup). When run with default windows encoding (windows-1250) there are no indexing errors, file content can be displayed, but filename gets garbled:
αφιλε.τχτ
- depends on
-
FE-7094 Add support for different file paths encoding for Mercurial
- Not Being Considered
- relates to
-
FE-6909 Indexing a Perforce repository fails with "P4ClientException - Unable to get file log" for files with non Unicode characters
- Closed
-
FE-7075 Support setting Mercurial repository's encoding in the FishEye config
- Closed
-
FE-6961 Add option to set default charset encoding for file content for Mercurial
- Not Being Considered