[hbm2hbmxml] problems with db-objects named "abc$ta_tab

rig · **Joined:** Wed Sep 06, 2006 9:54 am **Posts:** 9

Quote:

Hibernate version:
hibernate-3.1.3
HibernateTools-3.2.0.beta7, using the ant-tools!!!

Name and version of the database you are using:
Oracle9i 9.2.0.6.0

Hello out there,

i have an existing db-schema and named my db-objects like "abc$ta_tab_1", "abc$ta_tab_2", ... , "abc$ta_tab_n". first i do reverse engineering from the db with hbm2hbmxml. the generation of the hbm.xml-files works fine. but the generated xml contains some bugs.

in all generated "<many-to-one>"-relations the symbol "$" is replaced by the symbol ".". all other kind of generated references are ok.

so i have to edit the hbm.xml-files and replace the "." by "$". then i run "hbm2java". here, the generation of the java-code works find, too. but the java-code itselfs contains some bugs:

where the java-classes import other generated java-classes, the symbol "$" is replaces by the symbol ".", too.

a long long time ago, i did reverse engineering with middlegen for hibernate and hibernate2 based on the same db-schema and everything works fine. so i'm not sure, but is this a bug in hibernate-3.1.3 or did i some mistakes?

edit: the hibernate.reveng.xml-file i use contains only the following lines

Code:

<hibernate-reverse-engineering>
   <schema-selection match-schema="ABC_SCHEMA"/>
</hibernate-reverse-engineering>

best reguards
rig

max · **Posted:** Thu Sep 07, 2006 4:33 pm

submit some testcase describing it to jira (a *minimal* failing thing)

how did you rename the abc$ta_tab_1 in middlegen ? those names are not valid java...

rig · **Joined:** Wed Sep 06, 2006 9:54 am **Posts:** 9

max wrote:

those names are not valid java...

I use (generated with middlegen for hibernate2 and after changing the DOCTYPE in the hbm.xml-files from hibernate2 to hibernate3 currently running with hiberate-3.1.3) classnames like "pdc$taItem" or "pdc$taStatus" in Java (1.4.2 and 1.5) since more then one year and everything works fine. And from Sun's Java Tutorial clause "Naming":

Quote:

Variable names are case-sensitive. A variable's name can be any legal identifier — an unlimited-length sequence of Unicode letters and digits, beginning with a letter, the dollar sign "$", or the underscore character "_". The convention, however, is to always begin your variable names with a letter, not "$" or "_". Additionally, the dollar sign character, by convention, is never used at all. You may find some situations where auto-generated names will contain the dollar sign, but your variable names should always avoid using it. A similar convention exists for the underscore character; while it's technically legal to begin your variable's name with "_", this practice is discouraged. White space is not permitted.

so, using the "$" is not notallowed ("should be avoid" is not "must be avoid" in my mind and it's autogenerated)

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

max wrote:

how did you rename the abc$ta_tab_1 in middlegen

Take an editor (in this case "Kate") and replaced "." with "$" :-)

All jokes aside. I have e.g. 2 tables in my oracle-db: "PDC$TA_ITEM" and "PDC$TA_STATUS", with a foreign key references from "PDC$TA_ITEM" to "PDC$TA_STATUS". In my build.xml-File, I specified

Code:

The with hbm2hbmxml generated hbm.xml-Files named "pdc$taItem.hbm.xml" and "pdc$taStatus.hbm.xml". Because of the reference, the file "pdc$taItem.hbm.xml" contains amongst others following code:

Code:

<many-to-one name="taStatus" class="com.foo.bar.pdc.taStatus">
   <column name="STATUS_ID" />
</many-to-one>

Because i have a little bit notion of hbm.xml-files, i recognized, that com.foo.bar.pdc.taStatus are wrong. So I changed it to com.foo.bar.pdc$taStatus by hand, the spelling I know from middlegen for hibernate and run hbm2java. The ant-task generated all java-classes without errors. But when i opened the java-files with eclipse, I noticed that hbm2java genereted the following code in "taItem":

Code:

import com.foo.bar.pdc.taStatus;

When I change these to "import com.foo.bar.taStatus;, everything was fine in eclipse. But til now I had no opportunity presented itself for real-testing the generated javacode.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

At that time I started with hibernate2 and middlegen for hibernate, I did the same steps as now: existing db-schema -> hbm.xml-files -> java-classes. Unlike now the genereted hbm.xml-files called pdc_taItem.hbm.xml / pdc_taStatus.hbm.xml instead of pdc$taItem.hbm.xml / pdc$taStatus.hbm.xml today. And the generated JavaClasses called pdc$taItem.java / pdc$taStatus.java instead of taItem.java / taStatus.java today. When I don't want my JavaClasses named like the db-tables, I would specify an other name in the file "hibernate.reveng.xml".
In my mind as long as the names of the db-tables are valid JavaNames, they shouldn't be changed. If the names aren't valid JavaNames, throw an error while reverse engineering like "object name xyz is not a valid javaname. you have to specify a valid javaname in the hibernate.reveng.xml-file for these object."

best reguards
rig

max · **Posted:** Thu Sep 07, 2006 6:14 pm

put a *small* test/example for it in jira....but i don't promise anything since currently we have some code that strips out $ since it can be used to refer to inner classes in certain parts of the mapping files.

rig · **Joined:** Wed Sep 06, 2006 9:54 am **Posts:** 9

Hi,

max wrote:

put a *small* test/example for it in jira....but i don't promise anything since currently we have some code that strips out $ since it can be used to refer to inner classes in certain parts of the mapping files.

I have done so. Ticket-Number: HBX-751

reguards
rig