[sakai2-tcc] Mysql tablename case

[Matthew Jones]

Yea, Sam and I were considering the conversion effort, and if it could be
done with a regex or an XML parsing script it could be worth it, but I
think things like embedded hibernate queries with mixed case table names
referencing mixed case java classes will make that effort very difficult
and most likely time better spent elsewhere. [1] All the legacy stuff looks
uppercase, much of the newer and hibernate stuff is either lower or mixed.

Generally these were caught by someone early in the QA process (on a QA
server) when the script failed, and I'm not sure why it wasn't caught for
this cycle. In the past when I was at Michigan I tested all scripts (From
Sakai 2.6->2.8) on Linux and Windows in Mysql local to both as well as
Oracle. Now I just test on Mysql for Linux and this seems like this should
have came up. I know the number of active QA servers has dropped to really
just IU (Oracle/HSQL), Rutgers (Mysql) and Longsight (Mysql) and we have no
nightly server for Mysql that runs any conversions. We did run a 2.8->2.9
conversion test as well as Liquibase to compare differences prior to the
release. That work won't go away in either case fixed or not.

[1]
   <query name="findAllMembershipItemsForForumsForSite">
    <![CDATA[select membership from
org.sakaiproject.component.app.messageforums
.dao.hibernate.DBMembershipItemImpl as membership
            join membership.forum as forum
            join forum.area as area
            left join fetch membership.permissionLevel
            where
            area.id = :areaId
        ]]>
  </query>



On Fri, Jan 25, 2013 at 9:46 AM, Charles Severance <csev at umich.edu> wrote:

> I agree that there is no point in converting and code that does not match
> table or column names exactly is effectively introducing a bug on some
> fraction of Sakai production sites.
>
> /Chuck
>
>
> On Jan 25, 2013, at 9:39 AM, Matthew Buckett wrote:
>
> On Fri, Jan 25, 2013 at 11:51 AM, Aaron Zeckoski <azeckoski at unicon.net>
> wrote:
>
> I doubt this will elicit too much discussion but I think (maybe
>
> propose?) we should make it official that we require patches and
>
> code/scripts to use original tablename case (i.e. match what is
>
> already in the code). While many or most of us run with table name
>
> case sensitivity off in mysql, not everyone does and I believe it is
>
> still the default to have case sensitive table names. This has
>
> generally been the practice already for the CLE team but I am not sure
>
> it is written down anywhere.
>
>
> Yep, makes sense to me.
>
> - The option that hasn't been discussed is converting all the existing
> code to use lowercase table names and documenting the conversion
> process for exsiting installs doesn't seem worth the effort to me.
>
> --
>
>
>
> _______________________________________________
> sakai2-tcc mailing list
> sakai2-tcc at collab.sakaiproject.org
> http://collab.sakaiproject.org/mailman/listinfo/sakai2-tcc
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://collab.sakaiproject.org/pipermail/sakai2-tcc/attachments/20130125/fedd315a/attachment.html