Re: question for cvs2svn

On 06/24/2011 05:13 AM, hoyaojun wrote:
> 1. Based on my research, CVS RSC file will move in attic folder for recov=
er
> purpose(http://www.network-theory.co.uk/docs/cvsmanual/Attic.html). but w=
hat
> will happen in svn2cvs when converting this kind of file? the file will b=
e
> save in to the subversion backup system or it will just excluded in the
> converting? how i am going to put those file to subversion backup system?=
 =20

Attic files are handled correctly by cvs2svn: they are present in
exactly the part of history where they were present in CVS.

> 2. As the forum and FQA get telling on the optional on
> --retain-conflicting-attic-files to display on my attic folder.

You probably don't need the --retain-conflicting-attic-files option
unless you run into trouble with two RCS files with the same name, one
of which is in the attic and one of which is not in the attic.

> i using
> "cvs2svn --retain-conflicting-attic-files  -s <subversionNewRepositionPat=
h>
> <cvsRepositionPath>" and hit error as below:
>=20
> ----- pass 2 (CleanMetadataPass) -----
> Converting metadata to UTF8...
> WARNING: Problem decoding log message:
> -------------------------------------------------------------------------=
--
> add method(getPNBillDetail())=C2=A1=C2=A2modify code for support DF mode =
search and
> print DF agreement
>=20
> -------------------------------------------------------------------------=
--
> ERROR: There were warnings converting author names and/or log messages
> to Unicode (see messages above).  Please restart this pass
> with one or more '--encoding' parameters or with
> '--fallback-encoding'.
>=20
> As display, it suggest me to use --encoding=3DENC or --fallback-encoding=
=3DENC.
> when i read on the documentation, it show that log message will be lost w=
hen
> using --fallback-encoding which i not prefer on this. i using --encoding =
and=20
> it happen the same error(i'm tired on this with trying encoding in every
> "all language" in
> http://docs.python.org/library/codecs.html#standard-encodings). Is there =
any
> solution for this?

If you can figure out what encoding was used for the problematic log
message, then include it in an --encoding=3DENC option and all should be fi=
ne.

If you cannot figure out what encoding was used (or it could even be
that binary characters erroneously got into the log message), then there
is not much that can be done.  In this case you should use a
--fallback-encoding, which will decode what it can and replace any
unrecognized characters with "?" characters.

> 3. if i am not using optional --ratain-conflicting-attic-files and using
> suitable encoding code, it work. After everything had been converted, i
> check on the history using eclipse some of them had lost. In my cvs serve=
r
> most update history date will be 2/5/2011 but it showing the most latest =
is
> 2/7/2007. why this will be happen? is that bug in cvs2svn?or i need to ru=
n
> on cvs server in order to get everything correct? i try to run many time =
the
> result still is the same.=20

Obviously you have to use an up-to-date copy of the CVS repository for
the conversion.

Sometimes CVS dates are out of order, so it could be that the history is
correct but some commits are not where you expected to find them.  But
without more information it is impossible to guess.  The main goal of
cvs2svn is not to lose any information, so if you find that information
really IS lost, please report it as a bug.

Michael

--=20
Michael Haggerty
[email protected]
http://softwareswirl.blogspot.com/

------------------------------------------------------
http://cvs2svn.tigris.org/ds/viewMessage.do?dsForumId=3D1667&dsMessageId=3D=
2775567

To unsubscribe from this discussion, e-mail: [[email protected]=
s.org].