Re: [rdiff-backup-users] UnicodeEncodeError Mac OS X 10.5

[Max Duessar]

On Tue, Jul 22, 2008 at 11:51 AM, Andrew Ferguson <address@hidden> wrote:
> On Jul 21, 2008, at 3:55 AM, Max Duessar wrote:
>>
>> I've recently been getting these errors on my local backups (to USB
>> drive) on Mac OS X 10.5.4 using 1.1.16. I've upgraded to 1.1.17 and
>> the problem is the same. This is the error:
>>
>> Exception ''ascii' codec can't encode character u'\uf022' in position
>> 19: ordinal not in range(128)' raised of class '<type
>> 'exceptions.UnicodeEncodeError'>':
>>
>> File "/Library/Python/2.5/site-packages/rdiff_backup/eas_acls.py",
>> line 153, in EA2Record
>>  str_list.append('%s=0s%s' % (C.acl_quote(name), encoded_val))
>
>
> Hi Max,
>
> Thanks for your interesting error -- an extended attribute on one of your
> files has a Unicode character of 0xF022, which is well outside the ASCII
> range. Rdiff-backup is failing because the EA metadata file which it writes
> only expects ASCII characters.
>
> At this time, I have patched rdiff-backup in CVS so that it will skip over
> such an extended attribute. The correct fix would be to make the EA metadata
> file Unicode-aware, but as that may break the file format and we are close
> to a new stable release, I have postponed such a change. Another advantage
> to postponing such a change is that we will soon be able to move to Python
> 3000, which will be released in the fall with full Unicode support.
>
>
> best,
> Andrew
>

Thank you for the prompt patch to that issue; I hope to try it out
soon. However, I'm having some trouble compiling the CVS version. I
had previously installed rdiff-backup (and its dependencies) manually
without any problems and everything compiled fine. But now, with the
CVS version, I execute the "compilec.py" file in the rdiff-backup/
directory, and I get an error about incompatible architectures:

"ld: warning in /usr/local/lib/librsync.a, file is not of required architecture"

So I run this to see what architecture it's compiled for:

$ lipo -info /usr/local/lib/librsync.a
input file /usr/local/lib/librsync.a is not a fat file
Non-fat file: /usr/local/lib/librsync.a is architecture: i386

i386 seems fine after looking at the output for compilec.py showing (I
think) that it's using either i386 or ppc:

"gcc -Wl,-F. -bundle -undefined dynamic_lookup -arch i386 -arch ppc
build/temp.macosx-10.5-i386-2.5/_librsyncmodule.o -lrsync -o
build/lib.macosx-10.5-i386-2.5/_librsync.so"

It seems odd because I've compiled everything on this same machine
(MBP 15" Core2Duo), so everything should be compiled for the same
arch, right?

I've never compiled from CVS before so I can't say if this is a new
issue or not. I don't know if you've tried this on OS X before, but do
you have any ideas? (I can post the full output of the compilec.py if
needed, too).

--
Thanks,
Max