PAT 0.3 Released

[bugg_tb]

Its here at last, PAT 0.3 has arrived with glorious new features such as multiple tabbed queries and a more interactive scrolltable.

Be sure to head over to the download page and grab a copy and let us know what you think.

Tom

[pmalves]

Great news, congrats, everyone!!

In what version do you expect to have a drop-in replacement for jpivot?

[codek]

it looks amazing!

[gmoran]

Awesome news! I'll be poking around with it today

G

[dibe0015]

Hi,

the PAT standalone version wouldn't run again this time (on my Vista machine). This time I spent more time on the problem and it works if you change one value in the file
PentahoAnalysisTool\jetty\etc\jetty.xml

<Call name="addConnector">
<Arg>
<New class="org.mortbay.jetty.nio.SelectChannelConnector">
<Set name="host"><SystemProperty name="jetty.host" default="192.168.0.163"/></Set>
[...]

Set a default IP address in here (maybe just "localhost" would work too). Otherwise jetty willl listen on 0.0.0.0 on my machine.

Cheers,
Benjamin

[bugg_tb]

Hi Benjamin

Granted I know nothing about jetty, it just about holds together for this and will be sorted out properly at some point, interesting problem you face, I've tested it on a Windows box and a linux box and both have worked fine, are you trying a server client setup by any chance?

Thanks for delving.

Tom

[dibe0015]

Hi Tom,

so far I don't have any further plans with it. I just gave it a quick try, uploading a schema, using a xmla source (from Pentaho BI Server) and thats it. I could see the cubes and add Dimensions etc to a view but I wouldn't get any data to look at (I have to admit it is not a very stable schema I used). I don't have time during work to spend time for testing but I'll keep downloading every release thats getting published.

Benjamin

[bugg_tb]

Thats fine ben, any testing is fine by us, everyone has different requirements/data/schema etc so even if its just testing connectivity etc its still of huge benefit to us.

Hopefully soon we'll get the PAT plugin properly integrated so when y'all upgrade to BIServer-CITRUS you can all just plug it in to test it out.

Thanks again

Tom

[Peter Schmidt]

I tried running PAT, was able to connect to my test db using our current olap schema, but when I added a dimension and a fact, it never displayed any data. Any ideas?

[bugg_tb]

Rumour has it XMLA doesn't work, will be looked into tomorrow, normal mondrian connection `should` be ok.

[Peter Schmidt]

Perhaps I didn't use the right connection then,
I first uploaded my modrian schema and then used the following:
jdbc: postgresql://localhost:5432/dw_development
Was this not correct?
Thanks,
Peter

[bugg_tb]

Ah I may have misread your question slightly, once adding a dim and the measures there's a very cryptic bit you wont know about unless you tested halogen. Once you've dragged them into an axis, then you need to expand the + and right click on one of them, terrible usability, but its all we've got for the next month or so. If you're still not sure there's a small ogg you can grab on the downloads page showing you how it work.
Cheers for testing
Tom

[dibe0015]

Hi Tom,

the hint (right click) was helpful for me too

[bugg_tb]

lol, yeah sorry about that, it will be gone very soon, I promise.

[Peter Schmidt]

Thanks, the right-click did the trick.
You are right not very intuitive.

[bugg_tb]

Nope its a bag of sh*t.

Tom

[pstoellberger]

We know that there are still some major user handling problems in the current PAT release. 0.4 is the last release where we focus on the core of PAT so we can start with a fancy UI afterwards.

however the next release will bring some nice UI improvements as well, but some ugly things will need to remain for now.

we do take feedback serious and are glad that we're getting some finally ;-)

[Peter Schmidt]

Just some quick feedback off the top off my head then....

a. when I upload my mondrian schema and then connect, it takes awhile (30 sec) until I get available cubes
b. when I drag two dimensions over to a row, for example, I can't drag them to change the order
c. when right clicking (a user is not going to understand Member, Children, Include Children, Siblings, etcetera)
d. clicking on execute query model needs to provide feedback (is it running, was there something not configured to display output)?
e. as I think about it, isn't the drag and drop, 1st to Row/Column, then to query a little archaic? What about just dragging directly onto the canvas itself?
f. is the intention that this will replace jpivot in the BI Server?

Hope this is helpful,

Peter

[bugg_tb]

Just some quick feedback off the top off my head then....

An an equally quick retort...

a. when I upload my mondrian schema and then connect, it takes awhile (30 sec) until I get available cubes

I suspect thats mondrian trawling through all your schema, something to do with agg tables I believe, anyway technically not PAT's fault, although we need to be able to allow users to customize its internal mondrian instance.

b. when I drag two dimensions over to a row, for example, I can't drag them to change the order

Yup all this is temporary

c. when right clicking (a user is not going to understand Member, Children, Include Children, Siblings, etcetera)

This will be addressed, but I need to do some research into this as I don't really want to be making up words for them.

d. clicking on execute query model needs to provide feedback (is it running, was there something not configured to display output)?

Agreed

e. as I think about it, isn't the drag and drop, 1st to Row/Column, then to query a little archaic? What about just dragging directly onto the canvas itself?

Dragging to the table is planned for the future, check the mockups on the dev mailing list for the basic plan which we will be working towards.

f. is the intention that this will replace jpivot in the BI Server?

Initially to run along side certainly, Gretchen has already converted a 0.2 version into a Pentaho Plugin and whilst it needs a bit of work to upgrade we're aiming to have it sorted for when Citrus goes GA so you can all plug it in and test it.

Tom

[akrde]

Initially to run along side certainly, Gretchen has already converted a 0.2 version into a Pentaho Plugin and whilst it needs a bit of work to upgrade we're aiming to have it sorted for when Citrus goes GA so you can all plug it in and test it.

Sounds great, but what is a "GA" version?

[bugg_tb]

Tech speak, generally available, general availability that sorta thing

[akrde]

Tech speak, generally available, general availability that sorta thing

Ok, that describes what GA means, but what kind of release are GA releases: CI builds, Milestones, Release Candidates or Stable Builds?

[bugg_tb]

GA Is considered a stable build, for example on the pentaho download site you'll see M, Milestone builds, very experimental, RC, release candidates, nearly there, and GA or .stable, which are the stable versions.

Tom

[Peter Schmidt]

Tom,
Thanks for the quick replies.

a. when I upload my mondrian schema and then connect, it takes awhile (30 sec) until I get available cubes

I suspect thats mondrian trawling through all your schema, something to do with agg tables I believe, anyway technically not PAT's fault, although we need to be able to allow users to customize its internal mondrian instance.

Just an FYI, we don't have any aggr tables in the schema file.

Peter

[bugg_tb]

http://jira.pentaho.com/browse/MONDRIAN-543

See that would be my guess, but its only a guess

[pstoellberger]

Peter, what DB are you using? If you're using a schema-based DB how many tables do you have in there approximately?
How long does it take to open your schema in schema workbench? Same time? Did you set your DB connection within Schema workbench to a specific schema?

MONDRIAN-543 is only related to agg tables

we have already a JIRA about being able to configure the mondrian instance within PAT : http://jira.pentaho.com/browse/PAT-111
and also the "set to one specific schema" : http://jira.pentaho.com/browse/PAT-99

-paul

[Peter Schmidt]

Peter, what DB are you using?

postgres 8.3.7

If you're using a schema-based DB how many tables do you have in there approximately?

38 tables

How long does it take to open your schema in schema workbench? Same time? Did you set your DB connection within Schema workbench to a specific schema?

We don't use schema workbench when editing the mondrian schema file. We've found it is faster by hand and then you are also not limited to only the functionality that schema workbench allows you to edit/modify.

[pstoellberger]

ok i dont have a schema within reach to try it with that many tables

i can fully understand that schema workbench has these limitations. i just find it handy to start a new cube and do the tweaking manually.

could you try to load the schema in there anyway? just to have another application doing the same and compare the result... you could also try pentaho aggregation designer.. whatever loads the schema and scans it

thanks
-paul

[Peter Schmidt]

Here is what I saw:

When running Pentaho Aggregation Designer, I could configure the db connection, choose a mondrian schema file, click apply and in < 4 sec, I am able to select cubes from the drop down.
Then after selection the cube it took 50 sec (1st time) up to 2 minutes (2nd time) before I gained control again (ouch!).

The Pentaho Analysis Tool took 2 sec after clicking upload and 15 sec after clicking on connect. I
then created a very simple analysis view using the same cube above and it was around 30 sec.

So, I guess perhaps it wasn't taking all that long in PAT.

Peter

[pstoellberger]

Yeah it sounds like this is not really related to PAT but to Mondrian.
Thank you very much for the analysis/feedback.

If it keeps bugging you / you want us to not forget this issue feel free to file a JIRA bug report: http://jira.pentaho.com/browse/PAT

-paul