Hi,

Given the following column:

FDOC_TOTAL_AMT NUMBER(19,2)

Middlegen generates the following:

<property
name="fdocTotalAmount"
type="java.math.BigDecimal"
column="FDOC_TOTAL_AMT"
length="19"
/>

which results in ddl something like:

FDOC_TOTAL_AMT NUMBER(19, 19)

which is pretty useless :)

It should rather create something like this?

<property
name="fdocTotalAmount"
type="java.math.BigDecimal">
<column name="FDOC_TOTAL_AMT" sql-type="NUMERIC(19,2)"/>
</property>

Thanks,
Peter

True Middlegen does not generate the sql-type for the mapping documents. Maybe this should be an option (on a per-field basis).

A general question: How many users are using middlegen to generate mappings from which they use hbm2ddl to regenerate the schema?

Would it not be best to use the vendors tools to get the schema output. I can see a use for this if your changing database vendors or supporting multiple platforms. In that case you are (possibly) not working bottom up thus hand editing of the mapping documents is acceptable. How many people would be using one database vendors schema as the master then generating the schema for the other supported platforms? Interesting - I support multiple platforms in a current project myself but I am working from XDoclet as the master.

Hi David,

One more problem with XDoclet:

it doesn't export relations between a tables, like one-to-many, many-to-one... (I talk about hbm2ddl Ant task).

Is the a solution ?

--awt

You most certainly can define relationships using XDoclet which will generate the appropriate HBM mapping documents. This can then be used to generate the schema. In some cases you need to include additonal information in the HBMs (hence XDoclet) to provide preferences for DDL generation.

Maybe you should hand code a simple HBM example (parent - child) and get it to work first. See how the DDL is generated with various settings then you will know what need to be included at the XDoclet level.