p4convert: loading an SVN dump with high ascii codes in the file or directory name are not imported correctly.
converter.log displays the file names correctly, but the files are written and submitted with incorrect names. example svn dump, "example p4convert file naming issue.zip" loaded to upload.perforce.com containing sample svn dump, and resulting log files.
Context: This is an issue with converting on Windows to a non-unicode server in import mode.
Utf8 characters in filenames (like ae, 0xc3a6 ) appear to be read correctly and written corrrectly in the log, but
get translated to a windows-1252 character (0xe6) when written to the Perforce server.
I even used a file change watcher to monitor the working copy, and when p4convert creates files locally from the SVN dump, the file paths are still correct. These files are then added by p4java, and committed. So I suspect the bug is in p4java or p4d.
Context: This is an issue with converting on Windows to a non-unicode server in import mode.
Utf8 characters in filenames (like ae, 0xc3a6 ) appear to be read correctly and written corrrectly in the log, but
get translated to a windows-1252 character (0xe6) when written to the Perforce server.
dump file is in /warp/cases/00253048
I debugged this a bit, but I still don't understand it. I enabled logging with
And all the paths printed are correct.
I even used a file change watcher to monitor the working copy, and when p4convert creates files locally from the SVN dump, the file paths are still correct. These files are then added by p4java, and committed. So I suspect the bug is in p4java or p4d.