U++ forum: Welcome to the forum

Re: How to share a schema .sch with multiple header and source files [message #52211 is a reply to message #52193]

Fri, 09 August 2019 10:56

mirek
Messages: 14267
Registered: November 2005

Ultimate Member

Actually, SQL_Sqlite3 example has it already right:

This goes to header:

#define SCHEMADIALECT <plugin/sqlite3/Sqlite3Schema.h>
#define MODEL <SQL_Sqlite3/simple.sch>
#include "Sql/sch_header.h"

(hint, hint, see "_header" Smile

This goes to some C++ file:

#ifdef _DEBUG
#include <Sql/sch_schema.h>
#endif

#include <Sql/sch_source.h>

Now above provides two things. "_schema" part creates script generation routines that create sql scripts to actually create the schema in the database. Usually we are using script with "alter table add column" commands so that incremental changes can be added to database (we just ignore "thing already exists" errors when performing that script). In the example, this coresponds to

#ifdef _DEBUG
	SqlSchema sch(SQLITE3);
	All_Tables(sch);
	if(sch.ScriptChanged(SqlSchema::UPGRADE))
		SqlPerformScript(sch.Upgrade());
	if(sch.ScriptChanged(SqlSchema::ATTRIBUTES)) {
		SqlPerformScript(sch.Attributes());
	}
	if(sch.ScriptChanged(SqlSchema::CONFIG)) {
		SqlPerformScript(sch.ConfigDrop());
		SqlPerformScript(sch.Config());
	}
	sch.SaveNormal();
#endif

(SaveNormal saves scripts files to the disk, so that they can be used e.g. for the production machine).

Not that this is usually excluded in release (#ifdef _DEBUG) version of application... (but not necessary).

"_source" part creates code to support S_TABLE structures (e.g. mapping between column names and C++ variables) and is always needed.

Mirek

Report message to a moderator