empire-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rainer Döbele <doeb...@esteam.de>
Subject Code generator
Date Sat, 28 Nov 2009 22:35:24 GMT
Hi folks,

 

finally I have found some time to look at the code generator.

 

This time I decided to test with Microsoft SQL Server and I ran straight into a problem:

Unlike the Oracle driver I had before, I had to supply the schema name with the catalog parameter
of DatabaseMetaData.getTables().

Otherwise it wouldn't find the tables.

Looks like DatabaseMetaData.getTables() behaves very database specific.

Any idea how we can improve this?

 

Another issue is that the driver may create a sequence table to handle sequences.

There should be a way to ignore this sequence table.

I also had a problem with some character case issues on the column name.

This should work now.

 

Apart from that the general class layout looks good. 

I only felt the desire to rename the CodeGen class to CodeGenWriter. 

IMO this makes more sense. Hope nobody minds.

 

The most important thing missing are the relations.

I think they should be added next.

 

And I found that we should remove a few things from the generated database class.

The two generic methods are not specific to the database supplied an could go directly in
DBDatabase. 

I am however not sure, whether I really recommend to work this way.

This is more like traditional ORM's handle it and it breaks our paradigm - although it is
not wrong.

 

And the static accessor should go as well. We cannot decide whether someone wants one or multiple
instances and how they should be accessed. 

It is even likely that someone wants to derived a class from the generated database class
in order to be able to generate the classes again without loosing his modifications.

So if nobody objects I will remove this the next time.

 

Long time since we have last heard anything of Thomas Poling.

Maybe he has still not subscribed to empire-db-dev that why I put him on CC.

Thomas if you read this - give us a sign.

 

Benjamin IMO you have done a good job so far. 

I was surprised that splitting the process up in a parser and a writer by only using a DBDatabase
object has not been more difficult (or was it?).

And thanks Francis for checking everything in and fixing those "findbugs".

 

So how about doing the relations next?

Benjamin are you up for that? 

 

Regards

Rainer

 


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message