On Sat, Mar 9, 2019 at 11:10 AM Torsten Bögershausen <tboegi@xxxxxx> wrote: > > On Sat, Mar 09, 2019 at 09:36:34AM -0500, Jeffrey Walton wrote: > > > > I'm experiencing a failure in t0028-working-tree-encoding.sh. The > > first failure is test #3. The source states "source (test.utf16lebom, > > considered UTF-16LE-BOM)" but it looks like a UTF16-LE BOM followed by > > a UTF32-LE stream. > > > > Am I misunderstanding the data presentation? > > Thanks for the report. > > I think you understand it right. > > May be you can help us: Which OS are you using ? Fedora 29, x86_64 fully patched. However, I'm building Git and all of its dependencies with additional flags for testing. The prefix directory is /var/tmp and the lib directory is /var/tmp/lib64. RPATHS are set for everything being built, but I don't rule out those stupid path problems that plague Linux. In the past I have seen grep and awk from /bin use special builds of libraries in /var/tmp/lib64. I have not figured out how to tell programs in /bin to stop using test libraries in /var/tmp/lib64. > And what does > echo "hallo" | iconv -f UTF-8 -t UTF-16 | xxd > give ? $ PATH=/var/tmp/bin/:$PATH echo "hallo" | iconv -f UTF-8 -t UTF-16 | xxd 00000000: fffe 6800 6100 6c00 6c00 6f00 0a00 ..h.a.l.l.o... And: $ PATH=/var/tmp/bin/:$PATH echo "hallo" | /usr/bin/iconv -f UTF-8 -t UTF-16 | xxd 00000000: fffe 6800 6100 6c00 6c00 6f00 0a00 ..h.a.l.l.o... > We may need some more debugging, may be you can send the whole log file ? > Even if there is are a lot of ESC-sequences... Yes, absolutely. Which file would you like? (The only thing I can find is config.log). Jeff
