The file `masterChangelog.xml` contains a list of all changes (well, in fact more like a list of all changelog-files). The structure could look like this:
TIP: Keep in mind that the header-information can change, visit therefore the liquibase-website: http://www.liquibase.org Specific: the XML-format definition: http://www.liquibase.org/documentation/xml_format.html
So, thats where all our changes are located - but what about the changes _itself_?
As you can think of, they're stored in files located in the folders `struct` for DDL-stuff like _create_, _alter_ etc.
and `data` for DML-stuff like _insert_, _update_ etc.
If you want to know what to do add in these files take a look at the official liquibase documentation for futher information how to fill these XML files: http://www.liquibase.org/documentation/index.html
IMPORTANT: Always remember that you might have to define some rollback-entries. +
This is e.g. not needed for create-entries but for others like insert.
So, whats about that `liquibase.properties` file? The file is needed to specify some data here, e.g. the driver-name, classpath (of the driver), jdbc-url, etc.
You want to store these information there because otherwise you'd need to specify them on every call of liquibase. Not cool.
Now you're ready to run commands like `liquibase update` in the `PROJECTHOME/others/db_changes` folder.
=== in practice
.Let's say whe want to add a new table `FOOBAR` with some demo-data. We need to:
* create a new file in the struct-folder; e.g. `create_FOOBAR.xml`
* fill the file with changesets that will create a table
* add `include`-tag in the `masterChangelog.xml` for the created file
* create a new file in the data-foler; e.g. `demoData_FOOBAR.xml`
* fill the file with changesets that will insert data into the table and define *rollback-entries*
* add `include`-tag in the `masterChangelog.xml` for the created file
* to apply the changes simply run `liquibase update`
For example one featurepackage "add contact management" could contain several files itself:
----
create_org.xml
create_pers.xml
create_relation.xml
----
On the other hand, several featurepackages cannot be contained in one changelog-file.
== How to use it
Normally everything you need is provided by the project-template and the existing files in the git-repository.
Therefore it's easy to use and you can start immediately.
.Keep in mind:
* run `liquibase update` to apply possible changes after merging something into your repository
* apply modifications on the database always with liquibase; do not write a create-script in SQL. Write some changesets and execute them
* set correct `author`- and `id`-attributes in your changelog-files
== FAQ
[qanda]
What about unicode-columns like `NVARCHAR` or `NCLOB`?::
Simply define them in your changesets; This is something which is not completely verified:
Maybe columns like `NCLOB` cause problems in some DBMS like _MS SQL Server_
What if i swtich between branches and want to switch between different database-states?::
This is something that needs to be defined. Possible ways: rollback or recreate your database entirely
Is it possible to write an update-changeset where I can use a preparedStatement in the where-clause?::
This needs to be analysed and (the result) added here.
---
liquibase-readme
================
== Structure within ADITO
=== in theory
When using liquibase in an ADITO-System, your project structure _should_ look like this:
The file `masterChangelog.xml` contains a list of all changes (well, in fact more like a list of all changelog-files). The structure could look like this:
TIP: Keep in mind that the header-information can change, visit therefore the liquibase-website: http://www.liquibase.org Specific: the XML-format definition: http://www.liquibase.org/documentation/xml_format.html
So, thats where all our changes are located - but what about the changes _itself_?
As you can think of, they're stored in files located in the folders `struct` for DDL-stuff like _create_, _alter_ etc.
and `data` for DML-stuff like _insert_, _update_ etc.
If you want to know what to do add in these files take a look at the official liquibase documentation for futher information how to fill these XML files: http://www.liquibase.org/documentation/index.html
IMPORTANT: Always remember that you might have to define some rollback-entries. +
This is e.g. not needed for create-entries but for others like insert.
So, whats about that `liquibase.properties` file? The file is needed to specify some data here, e.g. the driver-name, classpath (of the driver), jdbc-url, etc.
You want to store these information there because otherwise you'd need to specify them on every call of liquibase. Not cool.
Now you're ready to run commands like `liquibase update` in the `PROJECTHOME/others/db_changes` folder.
=== in practice
.Let's say whe want to add a new table `FOOBAR` with some demo-data. We need to:
* create a new file in the struct-folder; e.g. `create_FOOBAR.xml`
* fill the file with changesets that will create a table
* add `include`-tag in the `masterChangelog.xml` for the created file
* create a new file in the data-foler; e.g. `demoData_FOOBAR.xml`
* fill the file with changesets that will insert data into the table and define *rollback-entries*
* add `include`-tag in the `masterChangelog.xml` for the created file
* to apply the changes simply run `liquibase update`
For example one featurepackage "add contact management" could contain several files itself:
----
create_org.xml
create_pers.xml
create_relation.xml
----
On the other hand, several featurepackages cannot be contained in one changelog-file.
== How to use it
Normally everything you need is provided by the project-template and the existing files in the git-repository.
Therefore it's easy to use and you can start immediately.
.Keep in mind:
* run `liquibase update` to apply possible changes after merging something into your repository
* apply modifications on the database always with liquibase; do not write a create-script in SQL. Write some changesets and execute them
* set correct `author`- and `id`-attributes in your changelog-files
== FAQ
[qanda]
What about unicode-columns like `NVARCHAR` or `NCLOB`?::
Simply define them in your changesets; This is something which is not completely verified:
Maybe columns like `NCLOB` cause problems in some DBMS like _MS SQL Server_
What if i swtich between branches and want to switch between different database-states?::
This is something that needs to be defined. Possible ways: rollback or recreate your database entirely
Is it possible to write an update-changeset where I can use a preparedStatement in the where-clause?::
This needs to be analysed and (the result) added here.
