Svoj,
Nice job on the patch. I approve the backup portions.
Chuck
> 2693 Sergey Vojtovich 2008-08-25
> WL#4515 - Falcon: fill INFORMATION_SCHEMA.FILES
>
> Fill I_S.FILES with information about Falcon data files. Remove
> I_S.FALCON_TABLESPACE_FILES. Make backup use I_S.FILES
> instead of
> I_S.FALCON_TABLESPACE_FILES.
...
> === modified file 'sql/handler.h'
> --- a/sql/handler.h 2008-08-07 12:58:45 +0000
> +++ b/sql/handler.h 2008-08-25 10:16:17 +0000
> @@ -612,8 +612,7 @@ enum enum_schema_tables
> SCH_TRIGGERS,
> SCH_USER_PRIVILEGES,
> SCH_VARIABLES,
> - SCH_VIEWS,
> - SCH_FALCON_TABLESPACE_FILES
> + SCH_VIEWS
> };
Just a style thing, but I would put SCH_FALCON_TABLESPACE_FILES last to keep
original values the same. No big deal.
...
> @@ -2490,6 +2474,7 @@ bool TablespaceObj::do_serialize(THD *th
> Materialize the serialization string.
>
> This method saves serialization string into a member variable.
> + Also extracts tablespace engine name from serialization string.
>
> @param[in] serialization_version version number of this
> interface
> @param[in] serialization the string from serialize()
> @@ -2505,6 +2490,11 @@ bool TablespaceObj::materialize(uint ser
> {
> DBUG_ENTER("TablespaceObj::materialize()");
> m_create_stmt.copy(*serialization);
> + /* Extract engine from create statement */
> + String tmp_str= String("=", 1, system_charset_info);
> + int pos= m_create_stmt.strrstr(tmp_str, m_create_stmt.length());
> + m_engine.copy(m_create_stmt.ptr() + pos + 1,
> m_create_stmt.length() - pos - 1,
> + system_charset_info);
This looks fine. It's the only way to get it on materialization. I do
something similar for checking DB on restore. What about case sensitivity?
Is there a difference between FALCON, falcon, Falcon, etc?
...
