Re: [gnome-db] Meta Store issue with MySQL.

[Bas Driessen <bas driessen xobas com>]

On Thu, 2009-07-30 at 16:17 +0200, Vivien Malerba wrote:
>     
>     
>     2009/7/30 Bas Driessen <bas driessen xobas com>
> >         
> >     
> >         > 2009/7/30 Bas Driessen <bas driessen xobas com>
> >         >
> >         >>  Hello,
> >         >>
> >         >> My database is in MySQL. When I run the following command to rebuild the
> >         >> MetaStore (which lives in a sqlite3 database):
> >         >>
> >         >> gda_connection_update_meta_store(connection, NULL, &error)
> >         >>
> >         >> I get the following error/warning:
> >         >>
> >         >> ** (process:22406): WARNING **: (+5): Wrong Holder value type, expected
> >         >> type 'boolean' when value's type is 'string' (Provider ThreadWrapper)
> >         >>
> >         >>
> >         >> and the MySQL table information is not in the MetaStore. In fact if I
> >         >> do:
> >         >>
> >         >> select * from _tables;
> >         >>
> >         >> nothing shows up.
> >         >>
> >         >> Anyone else having this problem?
> >         >>
> >         >
> >         > The reason for this error is that a SELECT statement internally run by the
> >         > MySQL provider to get meta data returned a column of type string instead
> >         > to
> >         > a boolean. To solve this, the correction consists in locating the faulty
> >         > SELECT, and use gda_connection_statement_execute_select_full() with an
> >         > array
> >         > of expected types for the columns instead of
> >         > gda_connection_statement_execute_select(). This is already done at several
> >         > places in gda-mysql-meta.c.
> >         >
> >         > If you want to try to correct it, and if you are in the master branch, you
> >         > can use the definitions in
> >         > libgda/providers-support/gda-meta-column-types.h
> >         > (which is dynamically generated).
> >         >
> >         > Otherwise, I'd need to have more information to correct the problem
> >         > myself.
> >         
> >         
> >     
> >         OK, I will try to locate and correct the issue in libgda. Just wanted to
> >         have confirmed first that this is an issue in libgda and not in my setup.
> >     
>     
>     To help you, you can uncomment the line 2324 of the gda-meta-store.c file, so you'll get debug info about all the results the provider is returning for the meta store.

OK, I am a step further in this (I think). In file providers/mysql/gda-mysql-meta.c, line 717 the structure should be changed from:

GType col_types_views[] = {
                G_TYPE_STRING, G_TYPE_STRING, G_TYPE_STRING, G_TYPE_STRING,
                G_TYPE_STRING, G_TYPE_STRING, G_TYPE_NONE
        };

to:


GType col_types_views[] = {
                G_TYPE_STRING, G_TYPE_STRING, G_TYPE_STRING, G_TYPE_STRING,
                G_TYPE_STRING, G_TYPE_BOOLEAN, G_TYPE_NONE
        };

The original error will be gone and it will continue past the point where it stopped. However, there are now many new error messages:


(process:8055): GLib-WARNING **: GError set over the top of a previous GError or uninitialized memory.
This indicates a bug in someone's code. You must ensure an error is NULL before it's set.
The overwriting error message was: Type gchararray not mapped for value 0

(process:8055): GLib-WARNING **: GError set over the top of a previous GError or uninitialized memory.
This indicates a bug in someone's code. You must ensure an error is NULL before it's set.
The overwriting error message was: Type gint not mapped for value (null)

(process:8055): GLib-WARNING **: GError set over the top of a previous GError or uninitialized memory.
This indicates a bug in someone's code. You must ensure an error is NULL before it's set.
The overwriting error message was: Invalid column bind data type. 253

(this repeats many many times)
============


(process:8055): GLib-WARNING **: GError set over the top of a previous GError or uninitialized memory.
This indicates a bug in someone's code. You must ensure an error is NULL before it's set.
The overwriting error message was: Unable to get value for row 0 and column 10

** (process:8055): WARNING **: Type gint not mapped for value (null) (Provider ThreadWrapper)

As a rollback is triggered, the meta store is empty.

I have no idea if I am pretty close to have this all working, or if this whole area is still up for a re-write/under development.

See also my previous thread for a reproducible case of this issue.

Any comments/tips/hints?

Thanks,
Bas.