Fwd: [Schevo-devel] New index structures -- important upgrade info

[Matthew Scott]

This is a good point. schevo/database/duru.py moved to schevo/database.py
and schevo/database/__init__.py moved to schevo/signal.py; the
schevo/database directory itself was removed.

So if anyone has a recent checkout that still has a schevo/database
directory, you should delete it as it is no longer part of the repository. I
tend to forget this because I very frequently run a script that removes
backups and *.pyc files.

Thanks, Joseph!


---------- Forwarded message ----------
From: joseph schlesinger <joes at alt2254.org>
Date: Sep 16, 2005 3:44 PM
Subject: Re: [Schevo-devel] New index structures -- important upgrade info
To: Matthew Scott <mscott at goldenspud.com>

<<off list to avoid the dissemination of possibly erroneous information>>

I *think* that if you run the scripts from the SVN tree as the docs
somewhere suggest, you have to remove the remnants of the
schevo/database because the svn update doesn't touch the .pyc files in
there.


Matthew Scott wrote:

> For those of you who are using the latest versions of Schevo, please
> note that a significant change in the internal structure of a Schevo
> database was made as of SVN revision 1011.
>
> If you have an existing Schevo database, you will need to have at
> least SVN revision 1016, which has the necessary upgrade code to
> remove old 'altkey' indices and create the new structures.
>
> To upgrade, BACK UP YOUR ORIGINAL DATABASE, then just use db_create as
> you did to create the database in the first place::
>
> db_create filename.durus schema.package
>
> Contrary to the db_create name, if a database already exists it will
> merely be updated based on the schema that you give.
>
> During the upgrade process you will see progress indicators as the new
> index structures are created.
>
> For more information about the new index structures and what they will
> be capable of, see http://docs.schevo.org/trunk/reference/indices.html
>
> Note: Specifying _index in a schema, using the 'by' method, and
> temporarily relaxing index uniqueness constraints are not yet
> implemented, but the new structures are able to support them when we
> do implement those features. Stay tuned! :)
>
> --
> Matthew R. Scott
>
>------------------------------------------------------------------------
>
>_______________________________________________
>Schevo-devel mailing list
>Schevo-devel at lists.orbtech.com
>http://lists.orbtech.com/mailman/listinfo/schevo-devel
>
>



--
Matthew R. Scott
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://schevo.org/lists/archives/schevo-devel/attachments/20050916/9c9505b6/attachment.htm