CVS migration help

Jelmer Vernooij jelmer at samba.org
Tue Oct 7 16:22:58 BST 2008 

Am Dienstag, den 07.10.2008, 17:03 +0200 schrieb Thomas Manson:
> Hi Michael,
>  
>   I've checkout the trunk version (the version on ubuntu hardy heron
> is quite old : 2.0.1)
>   succeed in cvs2svn conversion,
>  
>  unfortunately it crashes in the same way that bzr cvsps-import
> does : 
>  
>  
> thomas at home:~/temp/bzr$
> cat ../cvs2svn-tmp/git-blob.dat  ../cvs2svn-tmp/git-dump.dat |  bzr
> fast-import -
> bzr: ERROR: exceptions.UnicodeDecodeError: 'utf8' codec can't decode
> bytes in position 43-45: invalid data
The problem seems to be that one of the characters in your CVS
repository is not valid as UTF8 character. Did you specify the locale in
which the filenames are encoded explicitly somehow?

Git does not have this problem, since it does not interpret any of the
filenames you store in it. This has advantages (conversion can't fail
since you're not doing conversion at at all), but it also has
disadvantages - checking out the repository on hosts with a different
encoding breaks the filenames.

Cheers,

Jelmer

> Traceback (most recent call last):
>   File "/usr/lib/python2.5/site-packages/bzrlib/commands.py", line
> 834, in run_bzr_catch_errors
>     return run_bzr(argv)
>   File "/usr/lib/python2.5/site-packages/bzrlib/commands.py", line
> 790, in run_bzr
>     ret = run(*run_argv)
>   File "/usr/lib/python2.5/site-packages/bzrlib/commands.py", line
> 492, in run_argv_aliases
>     return self.run(**all_cmd_args)
>   File "/home/thomas/.bazaar/plugins/fastimport/__init__.py", line
> 199, in run
>     params, verbose)
>   File "/home/thomas/.bazaar/plugins/fastimport/__init__.py", line 77,
> in _run
>     return proc.process(p.iter_commands)
>   File "/home/thomas/.bazaar/plugins/fastimport/processor.py", line
> 83, in process
>     self._process(command_iter)
>   File
> "/home/thomas/.bazaar/plugins/fastimport/processors/generic_processor.py", line 317, in _process
>     processor.ImportProcessor._process(self, command_iter)
>   File "/home/thomas/.bazaar/plugins/fastimport/processor.py", line
> 105, in _process
>     handler(self, cmd)
>   File
> "/home/thomas/.bazaar/plugins/fastimport/processors/generic_processor.py", line 486, in commit_handler
>     handler.process()
>   File "/home/thomas/.bazaar/plugins/fastimport/processor.py", line
> 164, in process
>     for fc in self.command.file_iter():
>   File "/home/thomas/.bazaar/plugins/fastimport/parser.py", line 312,
> in iter_file_commands
>     yield self._parse_file_modify(line[2:])
>   File "/home/thomas/.bazaar/plugins/fastimport/parser.py", line 365,
> in _parse_file_modify
>     path = self._path(params[2])
>   File "/home/thomas/.bazaar/plugins/fastimport/parser.py", line 493,
> in _path
>     return s.decode('utf_8')
>   File "/usr/lib/python2.5/encodings/utf_8.py", line 16, in decode
>     return codecs.utf_8_decode(input, errors, True)
> UnicodeDecodeError: 'utf8' codec can't decode bytes in position 43-45:
> invalid data
> bzr 1.3.1 on python 2.5.2.final.0 (linux2)
> arguments: ['/usr/bin/bzr', 'fast-import', '-']
> encoding: 'UTF-8', fsenc: 'UTF-8', lang: 'en_US.UTF-8'
> plugins:
> 
> bzrtools             /usr/lib/python2.5/site-packages/bzrlib/plugins/bzrtools [1.3.0]
>   cvsps_import         /home/thomas/.bazaar/plugins/cvsps_import
> [unknown]
>   fastimport           /home/thomas/.bazaar/plugins/fastimport
> [unknown]
> 
> launchpad            /usr/lib/python2.5/site-packages/bzrlib/plugins/launchpad [unknown]
> *** Bazaar has encountered an internal error.
>     Please report a bug at https://bugs.launchpad.net/bzr/+filebug
>     including this traceback, and a description of what you
>     were doing when the error occurred.
> 
>  
> I don't think it's related to cvs2svn or cvsps as it fails in both
> cases.
> It should be a bzr bug.
>  
> I've successfully converted my project to git repository format with
> these set of command : 
>  
> export CVSROOT=/home/thomas/temp/cvs2git/cvs/files
> 
> git cvsimport -C /home/thomas/temp/cvs2gitOutput/crf-irp
> crf-irp
> git cvsimport -C /home/thomas/temp/cvs2gitOutput/crf-irp-model
> crf-irp-model
> git cvsimport -C /home/thomas/temp/cvs2gitOutput/crf-irp-monitor
> crf-irp-monitor
> git cvsimport -C /home/thomas/temp/cvs2gitOutput/crf-irp-portail
> crf-irp-portail
> git cvsimport -C /home/thomas/temp/cvs2gitOutput/crf-irp-utilities
> crf-irp-utilities
>  
>  
> Is it possible to convert the git version of my sources to bzr ? maybe
> it would be successfull.
>  
>  
> Thomas.
>  
>   
> 
> 
> On Tue, Oct 7, 2008 at 12:41, Michael Haggerty <mhagger at alum.mit.edu>
> wrote:
>         Jelmer Vernooij wrote:
>         > Am Dienstag, den 07.10.2008, 00:01 +0200 schrieb Thomas
>         Manson:
>         >> I've look to it... but didn't tryed yet...
>         >>
>         >> It really misses straightforward howto (for all tools
>         except bzr
>         >> cvsimport)
>         > cvsps-import should be the best solution here, we should
>         just fixing
>         > that imho. What's blocking you from using it?
>         
>         
>         No conversion tool that is based on cvsps will be able to do a
>         truly
>         reliable job of migrating from CVS.  cvsps, which was written
>         for
>         another purpose, simply is not robust enough and does not emit
>         enough
>         information for a complete conversion.  I gave many concrete
>         examples of
>         its shortcomings on the Mercurial mailing list [1].
>         
>         Deducing a project's history from CVS's incomplete records is
>         a very
>         tricky thing; cvs2svn's feature list [2] will give you an idea
>         of the
>         kinds of things an industrial-strength converter needs to
>         handle.
>         cvs2svn deduces the CVS changesets itself, using a much more
>         robust
>         algorithm than that used by cvsps.  (The main disadvantage of
>         cvs2svn is
>         that it can only be used for one-time conversions, not for
>         tracking a
>         live CVS repository incrementally.)
>         
>         cvs2svn/cvs2git can create output in git-fast-import format
>         [3], which
>         should also be readable by the bzr fast-import tool.  It
>         hasn't gotten
>         much testing in "cvs2bzr" mode, but given that 90% of the job
>         is
>         inferring CVS's history, it should not be too much work to fix
>         any
>         problems in the "2bzr" part.  Therefore, any feedback would be
>         much
>         appreciated.
>         
>         (By the way, if you want to use cvs2svn to convert to bzr, I
>         suggest
>         that you use the trunk version of cvs2svn, which has several
>         improvements compared to release 2.1.1.)
>         
>         Michael
>         
>         [1]
>         http://selenic.com/pipermail/mercurial-devel/2008-February/004975.html
>         
>         [2] http://cvs2svn.tigris.org/features.html
>         
>         [3] http://cvs2svn.tigris.org/cvs2git.html
> 

-- 
Jelmer Vernooij <jelmer at samba.org> - http://samba.org/~jelmer/
Jabber: jelmer at jabber.fsfe.org

More information about the bazaar mailing list