Somehow i feel bad with spoon! So suit up.

[ferhatsb]

Hey kettles,

I have lead a migration project of a CMS application about 2 months ago and kettle helped us a lot. However that command line arguments and the way of configuring kettle via command line (refering to spoon) was not fulfilling our demands.

Instead of that, i have developed some wrapper main application that can be configured via xml file and handles multiple .ktr files, there are more things need to be work on but for me it is much better way and more flexible.

I will blog it soon but lets say it is at alpha release and currently configuration xml looks like,

<?xml version="1.0" encoding="UTF-8"?>
<configuration>

<sourcedbhostname>host db ip</sourcedbhostname>
<sourcedbport>port number</sourcedbport>
<sourcedbname>db name</sourcedbname>
<sourcedbusername>user</sourcedbusername>
<sourcedbpassword>password</sourcedbpassword>


<destinationdbhostname></destinationdbhostname>
<destinationdbport></destinationdbport>
<destinationdbname></destinationdbname>
<destinationdbusername></destinationdbusername>
<destinationdbpassword></destinationdbpassword>

<exportpath>/home/ferhat/fatwire/export/xml</exportpath>
<importpath></importpath>
<commitsize>100</commitsize>

<transformations>
<transformation>foo.ktr</transformation>
<transformation>foo1.ktr</transformation>
<transformation>foo2.ktr</transformation>
</transformations>
</configuration>

Do you feel comfortable with configuration approach of spoon?
Do you think my application can be usefull?

Would be glad to hear your comments,

KR,
Ferhat.

[MattCasters]

Hi Ferhat,

I'm not quite sure what problem this is supposed to solve.
What exactly is your problem? These variables can already be set in kettle.properties. So perhaps it's something else?

Feel free to share the experience. As Dr Indiana Jones once said: "Nothing shocks me, I'm a scientist" ;-)

Matt

[ferhatsb]

Hey Matt,

Sorry for miss type, was talking about pan not spoon,

well;

For my case, i had to deliver an application to infrastructure team based on another country for using migrate operations. The requirements were;

Properties should be easily configurable.
All transformations should run atleast once in an order together.
If we look at that line;

pan.sh -file="/PRD/Customer Dimension.ktr" -level=Minimal

doing it with command line arguments of course was an option,
but that required lots of shell codes, which is hard to read,
configure and less flexible.

What i'm trying to do is or i have done some of it, passing
arguments to pan via a xml file, that can also run multiple ktrs
in an order.

Perhaps i have reinvent the wheel, i'm not sure, what you think?

KR,
Ferhat.

[MattCasters]

Hi Ferhat,

Up to a certain extent you do have a point.
If all you want to do is populate a single dimension, then it becomes a little bit tedious to set all the required parameters if you don't want to use kettle.properties with standard values per system.

However please keep in mind, that this is a highly un-usual situation.
Usually people populate a complete warehouse in one go.

Don't forget it is possible to read whatever format of file you have (including yours) in a job and set the variables accordingly.

All that being said, I'm not against creating an execution configuration option for transformations and jobs. The XML format is already defined, all we need is a parameter to support it for pan/kitchen.

Why don't you file a JIRA feature request and I'll keep an eye on it for PDI4 :-)

Take care,

Matt

[ferhatsb]

Hi Matt,

PDI-2846 is up and waiting for your attention,

KR,
Ferhat.