I have not seen that error before. Is it a problem with one of the
double quotes being a different character than the normal that the XML
parser understands?
Nathan
-Original Message-
From: jhavero [mailto:jhav...@gmail.com]
Sent: Monday, January 12, 2009 5:07 PM
To:
Good point. Probably best to still run it against your real database
type before a release.
I did get a version of my test data running on h2 but found them to
run in the same time it took to run them aginst mysql. It seemed
surprising. I'll keep looking into it.
Nathan
On Jan 16,
We don't have anyone working on a database-specific datatype to
general datatype converter currently.
I'm not sure if it could be done in general fasion since there are so
many variations.
You could write a custom database subclass to convert the types the
way it would make sense for
Is your changelog file in a jar or a standard directory ?
Nathan
On Jan 31, 2009, at 1:35 AM, MichaelM t3nq9n...@sneakemail.com
wrote:
I am running on Windows. Tried various ways of specifying the
includepath,
with no result.
I'm not sure what you mean by configuration of the base
I added a link to a Best Practices page on the site's sidebar that had
been built up a couple weeks ago but never actually liked to.
http://www.liquibase.org/bestpractices
Anything you or anyone else think should be documented in there can be.
To answer some of your questions that I don't think
If you are looking to reuse files that you could load with the oracle loader,
you could use the executeShellCommand tag to make a call to the oracle loader
and have it run as it used to. That may or may not be of help to you depending
on your setup and what you are trying to accomplish.
It should just be doing that if you are doing a diff. Just making sure
that is what you are doing and it is not making those calls on its
own...
I checked and it looks like the call to the 1 = 0 statement is trying to
escape the table name and the postgres Database class is coded to quote
them
If you run liquibase --help it will give you the options available from the
command line. There should be a --diffTypes option or something similar.
Nathan
On Feb 25, 2009, at 6:11 PM, Pradnya Gawade pradnya.gawa...@gmail.com wrote:
Thanks Ben for the quick reply.
I think you are forgetting to wrap your sql tag in a changeSet tag.
You can support multi-line sql by using the splitStatements=false
attribute
Nathan
From: Pradnya Gawade [mailto:pradnya.gawa...@gmail.com]
Sent: Thursday, February 26, 2009 5:07 PM
To:
)
at liquibase.commandline.Main.main(Main.java:95)
I don't know what is wrong in xml syntax. Please help. Please provide
links to some sample sql tags if you have.
Thanks and Regards,
Pradnya
On Thu, Feb 26, 2009 at 6:15 PM, Voxland, Nathan
nvoxl...@intelligentinsites.com wrote:
I think you
Or are you wondering of you can build up an in memory database schema
and then apply it when it is done?
That would not be possible either since we rely on the database to
supply the state, not an in memory representation.
Your only option with the IDE would be to build your database up
Is that a feature others would like?
My assumption has been that it doesn't really hurt to always force it,
and then you know things will be working right.
Nathan
On Feb 27, 2009, at 8:43 PM, Paul Keeble cs...@yahoo.co.uk wrote:
There is currently no way to turn it off.
Paul
From:
approach to data
modelling (and I thought that I could leverage liquibase for that
purpose). I'm wondering if I'm trying to overtwist it's core purpose of
managing schema changes.
Dave
From: Voxland, Nathan [mailto:nvoxl...@intelligentinsites.com]
Sent
I person can dream, can't they? :)
I'll add it as a feature request.
Nathan
-Original Message-
From: Marc Logemann [mailto:l...@logemann.org]
Sent: Monday, March 02, 2009 10:45 AM
To: liquibase-user@lists.sourceforge.net
Subject: [Liquibase-user] SpringLiquibase.html with completely
I think the easiest way is to use the parameter functionality.
http://www.liquibase.org/manual/changelog_parameters
You can make your changeset as schemaName=${schema} and then pass the
schema you want into liquibase.
Nathan
From: Pradnya Gawade [mailto:pradnya.gawa...@gmail.com]
It should be outputting both an add primary key statement and a REORG
statement. From what I could tell, DB2 liked to/needed to have reorg
called after an add primary key.
Is the add primary key statement not being sent? You are running maven,
correct?
Nathan
From: Thanh Nguyen
There is not support for either case with the normal tags, you are
going to have to use sql tags. Inserting data via sequence is
something we may support with the insert tag eventually since it is a
cross-database use case. Creating synonyms is oracle specific so it
will probably always need to
No. I can add in a check and only output the reorg command with v9 or
greater DB2, but that won't make it until the next release.
Is there any other reorg-like statements you need to make with 8.2?
Nathan
-Original Message-
From: tomnguyen [mailto:tomn...@yahoo.com]
Sent: Friday,
A good example of duplicate ids/authors is if you use an incrementing
number for your ids. Each changelog would have changeset ids 1,2,3,4
etc. and so they wouldn't be unique across all changelogs. Adding the
filename and path as part of the identifier makes it so you can use that
id scheme (and
I was recently able to write a script to take a mysql dumpfile and
load it into an h2 in-memory database. I have found h2 to be
considerably faster than hsql in the liquibase unit tests.
We are building up a test dataset in mysql with a good variety of
normal and edge case data that we can
I have thought about dbunit in the past, but have always been
concerned about changes to the database scema breaking your dbunit
loading tests. Is there a way to avoid that or is it not really a
problem. I have generally used liquibase with contexts for that
reason rather than dbunit.
for the input.
Dave
Voxland, Nathan wrote:
I was recently able to write a script to take a mysql dumpfile and
load it into an h2 in-memory database. I have found h2 to be
considerably faster than hsql in the liquibase unit tests.
We are building up a test dataset in mysql with a good variety
A couple options:
- You could create a subclass of the MySqlDatabase class and override
the getClobType() method. You can pass the databaseClassName into
liquibase.
- You could also specify the type as a parameter: column
type=${clobType}/ and then pass the value into liquibase.
I would think
That does work well if you are not wanting to change your database
default.
For 1.10 I am thinking of focusing on extensibility and including a
contrib or plugin area to the site to include things like this.
Nathan
On Mar 29, 2009, at 12:45 PM, Jorge Falcão jlbfal...@gmail.com wrote:
I
That would be a very useful feature, but currently liquibase does not
support any sort of direct support for parallel tasks. Plus, like you
said, you don't want the liquibase process to be hanging round for
possibly days.
I'm not sure how you run your jobs, however. Once concern I have is if
There is a database diff tool as part of liquibase that you can use to
generate reports or use programmatically to determine (at least some)
differences between databases. You would need a standard database to
compare to because we cannot diff a database with a changelog.
You could write a
I don't think that executing a large number of changes is going to
corrupt your database metadata, but performance begins to degrade over
time.
Unfortunately there is not a great automated way to combine changesets.
There is the diff tool (http://www.liquibase.org/manual/diff) which can
, if there are
changes after changeSet #2, would they also fail when #2 hadn't been
applied? Would every changeSet need to include the sqlCheck or it
built-in that changeSet M can't be applied until changeSet N is applied
(M N)?
From: Voxland, Nathan
There is a dropUniqueConstraint tag, but it does require the
constraint name because that is how most database (including hsql)
structure their sql. HSQL is ALTER TABLE tablename DROP CONSTRAINT
constraintname
Without a constraint name, I don't think there is much you can do. I
would say you
these tables?
Thanx for the CREATEs. Should I log a jira somewhere?
Cheers
Sven
On Fri, Apr 3, 2009 at 3:19 PM, Voxland, Nathan nvoxl...@intelligentinsites.com
wrote:
Liquibase should add create table commands for the databasechangelog
and databasechangeloglock tables in your script
LiquiBase 1.9.2 has been released. It is a bugfix release, mainly
around issues with the includeAll tag.
Like always, it can be downloaded from http://www.liquibase.org/download
Nathan
--
This SF.net email is
I've used this plugin a bit in the past, and I never got around to
mentioning that I thought it worked great. When I was working on the
IDE plug-ins for intellij, I had always been approaching it from a
database-centric view where I was trying to create something let you
make changes to your
We have been providing 1.4 support via retroweaver so far.
Is anyone actually using using Java 1.4 or should we not bother with it
anymore? I am planning on doing some refactoring with the 1.10 release
and it would be nice to not have to worry about 1.4 support, but if
there is demand for
There is a nested changeLogProperty tag you can use.
Try:
updateDatabase
changeLogFile=${persistency.option.db.changelog.file}
driver=${persistency.option.database.driver}
url=${persistency.option.database.url}
I'll take a look at it. I think I had ran into it in testing, but
thought it looked database-specific.
What database are you using?
Nathan
From: Animesh Jain [mailto:anim...@itasveer.com]
Sent: Wednesday, April 08, 2009 8:49 AM
To: liquibase-user@lists.sourceforge.net
Subject:
Subject: Re: [Liquibase-user] Liquibase 1.9.2 generateChangelog Ant
targetiscreating createIndex elements without the column namewhichis
required
I'm trying on MySQL 5.0.51 on a windows machine
Thanks
Animesh
On Wed, Apr 8, 2009 at 10:05 PM, Voxland, Nathan
nvoxl...@intelligentinsites.com wrote
The way I understand it, you guys are looking for a way to run updateSQL, but
not pass it a database connection to check the databasechangelog table of, but
instead output all SQL that would be run against an empty databasechangelog
tablem with each changeset block having an “if row exists”
codebase.
Animesh
On Thu, Apr 9, 2009 at 12:48 AM, Voxland, Nathan
nvoxl...@intelligentinsites.com wrote:
I uploaded a test version that should fix the problem. Could you double
check that it works for you?
http://www.liquibase.org/liquibase-SNAPSHOT.jar
Nathan
From: Animesh Jain [mailto:anim
It was because our diff comparison for missing columns was not checking
for nullability. I put a new snapshot up at
http://www.liquibase.org/liquibase-SNAPSHOT.jar see if that is better.
Nathan
From: Animesh Jain [mailto:anim...@itasveer.com]
Sent: Thursday, April 09, 2009 2:15 AM
To:
:)
Animesh
On Fri, Apr 10, 2009 at 1:37 AM, Voxland, Nathan nvoxl...@intelligentinsites.com
wrote:
It was because our diff comparison for missing columns was not
checking for nullability. I put a new snapshot up at http://www.liquibase.org/liquibase-SNAPSHOT.jar
see if that is better
There is a logicalFilePath attribute on the root databaseChangeLog for
just this usage. If you need to move your changelog to another physical
location, you set the logicalFilePath to what it used to be and that is
the path that will be used to compare and store in the databasechangelog
table.
The reason we have the changelog path as part of the unique identifier
for a changeset is so you do not need to create an ID naming strategy
that is unique across all changelogs. For example, I just use a
sequential number that starts at 1 for each new changelog and they are
still uniquely
Close on the heels of 1.9.2, 1.9.3 is out which fixes an issue with the
generation of XML for createIndex statements. This may only affect
users of the generateChangeLog statement, but if you are getting MD5Sum
validation errors on your createIndex changes, upgrading to this version
may resolve
There is a createNode method on ChangeSet that will return an XML Node
object that you can append to your Document. When your changelog
Document is complete, there is a liquibase.xml.DefaultXmlWriter
convenience class that will output your document into an outputstream.
Nathan
-Original
is actually being sent through?
Nathan
-Original Message-
From: Voxland, Nathan [mailto:nvoxl...@intelligentinsites.com]
Sent: Thursday, April 23, 2009 9:43 AM
To: liquibase-user@lists.sourceforge.net
Subject: Re: [Liquibase-user] Error generating changelog
Sorry for the delay answering
It is not currently, ou'll have to specify your createTable statement
using sql as well.
Can you set utf8 as the default in your database config?
Nathan
From: Sven Haiges [mailto:sven.hai...@googlemail.com]
Sent: Thursday, April 23, 2009 1:20 PM
To:
There is no way to update to a particular changeset. This is partly by
design, because we don't want you to think you are creating particular
database versions with each changeset, but rather listing changes that
need to be applied to get to the current state. Especially when you get
into
Voxland, Nathan wrote:
There is no way to update to a particular changeset. This is partly
by
design, because we don't want you to think you are creating particular
database versions with each changeset, but rather listing changes
that
need to be applied to get to the current state. Especially
not a
development time issue, unfortunately.
With regards to adding that command, do you need me to JIRA it?
- Darryl
Voxland, Nathan wrote:
It should not be difficult to add, I can look at adding it for 1.10.
You may be able to do what you are wanting using preconditions.
You can
create
The dots in the properties may be throwing it off. What if you try it
with underscores?
Nathan
On Apr 27, 2009, at 4:28 PM, jfrapper jfrap...@lycos.com wrote:
Greetings,
I am using changelog parameters in my liquibase changelogs.
Everything
works fine from the ant task, but I cant get
The trouble is that not all databases return really useful comments for
tables, so we didn't want to automatically move them along to the new
database. For example, the comments for mysql tables are usually
something like InnoDB FREE 53k. It is something that we would like to
work around and get
defaultValueDate=sysdate name=UPDATED_DATE
type=TIMESTAMP(6)/
/createTable
/changeSet
Again, it works fine from ant, but not the command line.
Cheers,
JF
Voxland, Nathan wrote:
The dots in the properties may be throwing it off. What if you try it
with underscores?
Nathan
I have set up a LiquiBase Community Forum at http://www.liquibase.org/forum
that will take the place of the mailing lists.
My hope is that it will lead to a lower bar for entry to ask or answer
questions, allow better searching of past messages, and more.
Please check it out, register
by clicking the Notify button on the board. I would
recommend that for the good old liquibase-user fans like me...
Diego
On Wed, Apr 29, 2009 at 5:15 AM, Voxland, Nathan
nvoxl...@intelligentinsites.com wrote:
I have set up a LiquiBase Community Forum at
http://www.liquibase.org/forum
that will take
The built jar must not be quite right. I think I had tried pulling out only
the core library source, not the IDE code etc. and must be missing some
things that are needed.
I would mention that the trunk is significantly different and probably broken
compared to 1.9.3, so you would probably
a note to do so
though.
:-)
Voxland, Nathan wrote:
If you could add it to the sourceforge tracker, that would be great.
Let us know if you have questions with the preconditions.
Nathan
On Apr 24, 2009, at 11:44 AM, Darryl Pentz djpe...@yahoo.com
wrote:
Thanks Nathan, I'll take
You forgot to include the .java file, but I'll just apply what I had
sent and take a look.
Nathan
-Original Message-
From: jfrapper [mailto:jfrap...@lycos.com]
Sent: Thursday, April 30, 2009 9:18 AM
To: liquibase-user@lists.sourceforge.net
Subject: Re: [Liquibase-user] Changelog
/liquibase-SNAPSHOT.jar that has the
fix. Take a look and let me know how it works for you.
Nathan
-Original Message-
From: Voxland, Nathan [mailto:nvoxl...@intelligentinsites.com]
Sent: Thursday, April 30, 2009 9:57 AM
To: liquibase-user@lists.sourceforge.net
Subject: Re: [Liquibase-user
OK, I'll take a look at the .bat when I get a chance then to try to
figure out where it is going wrong.
Thanks for the troubleshooting help.
Nathan
-Original Message-
From: jfrapper [mailto:jfrap...@lycos.com]
Sent: Monday, May 04, 2009 12:26 PM
To: liquibase-user@lists.sourceforge.net
The reason we go through the URI/URL conversions is partly limitations
on the built-in java apis and partly to support pulling from
non-filesystem based locations (jar files, files over the network).
I'll apply your fix and do some more testing on it. Thanks a lot for
the fix.
Nathan
I think the problem is that the oracle JDBC driver does not have all the
capabilities of sqlplus. You will probably need to use the execute
shell command (http://www.liquibase.org/manual/execute_shell_command)
tag to call sqlplus and pass it the sql to run. Otherwise, you will
have to convert
The upcoming 2.0 version will allow much better support for integrating
features like this into the liquibase life cycle. For now, the best
option may be to create a custom change class
(http://www.liquibase.org/manual/custom_refactoring_class) that wraps
the generation of your exec statements
Thanks for the reproduction information. I'll take a look at it.
Nathan
-Original Message-
From: CarloV [mailto:carlo.verdecc...@scs-computers.com]
Sent: Thursday, May 21, 2009 2:02 AM
To: liquibase-user@lists.sourceforge.net
Subject: Re: [Liquibase-user] Diff finds foreign key
I did a couple related blog posts a while ago at
http://blog.liquibase.org/2007/06/building-database-tests-that-dont-brea
k.html.
I have always used liquibase both liquibase and mysqldump for managing
my test datasets, and both work well. Keeping it all in liquibase
changelogs makes it easy
22:40 schrieb Voxland, Nathan:
If you are interested in adding support, that would be great. The
easiest way is to register on liquibase.jira.com, check out the code
using your username/password, and commit what you need to change. We
are
experimenting with
The state of the code right
17:24 schrieb Voxland, Nathan:
Other databases have that problem too. I think the issue is actually
that our PK names are too long and db2 is truncating it down to a
non-unique version. Normally, I think it is PK_DATABASECHANGELOG and
PK_DATABASECHANGELOGLOCK
Nathan
-Original Message
Is the main site down for you or is it just bad links?
I just checked and it was up for me.
Nathan
On Jun 25, 2009, at 9:25 AM, Robert Fischer robert.fisc...@smokejumperit.com
wrote:
Trying to get to it to pull down the most recent JAR, and it's not
loading.
http://downforeveryone.com
67 matches
Mail list logo
