Business Objects to Join Eclipse Foundation

daria42 writes "Business intelligence specialist Business Objects is the latest software maker to join the Eclipse Foundation, and says it will move several products onto the open source platform -- but it's not yet saying which. 'We won't fight it, we'll embrace it,' said one of the company's executives in Sydney last week, talking about the open source software model. 'One of the reasons we've chosen to go with the Eclipse platform, rather than any of the other open source types,' she said, 'is that [Eclipse] actually has a model where vendors can sell value-added products into it, but still provide the service components.'"

Comments

So, enlighten me.

[TinBromide]

It looks to me like They're joining an initiative for opensource objects that offer a bit of open source but allow the real meat of thier offerings to be pay? So, adobe has been doing this for as long as i can remember with thier pdf reader, why didn't business objects join adobe :)

Business Objects?

[RandoX]

Which business? And why are they objecting? Seriously, there should be at least SOME details in the synopsis.

Interesting, considering...

[kat11v]

Interesting considering that right now they are currently hiring/hired three new patent agents (my friend in Vancouver being one of them).

Then again maybe it's just my slightly paranoid conscience jumping to silly conclusions.

I tried to join the Eclipse Foundation...

[digitaldc]

...but I was overshadowed by a brighter Java developer.

Totally off topic old joke

[Flying pig]

The story reminds me of the old Naval urban legend of the file being passed around a shore establishment and getting fatter and fatter without getting closer to completion until somebody scrawls on the front page "Round objects". Eventually the file returns but now on the front page a senior officer has written "Please will Mr. Round state his rank and clarify his objection".

So I have to ask: What is it about business that it objects to Eclipse?

Going out on a limb...

[aicrules]

But much like most other corporate entities that do this, while we don't know which products they will release open soruce, I'm guessing it won't be the one that they typically charge six figures for in license fees.

I do like Business Objects though, it's a decent platform.

Re:Repeat after me...

[ponds]

I think by the "platform" they mean eclipse, and the submitter is just making sure that we know that it's open source. Consider the quote right after it that says "the Eclipse platform."

Crystal Reports for Java?

[RingDev]

Oh my god do I feel sorry for you. Custom reporting solutions have often been descriped as a pile of crap. Business Objects, the current owners of Crystal Reports (Formerly of Seagate, Crystal Decision, and lord knows what other companies) has long been considered the best of these options. In otherwords, they were/are the best piece of crap solution available. Microsoft (insert booing and hissing here) has finally decided to create their own reporting solution (SQL Reporting) which from what I've heard is a significant improvement over CR XI (yes, 11 versions and there are still limitations on Cut and Paste)

-Rick

Re:Crystal Reports for Java?

[LDoggg_]

I haven't used it, but Cyrstal reports for Java has been around for a while. A demo of it came bundled with the last few version of jbuilder I've used.

Re:Crystal Reports for Java?

[kpharmer]

> Oh my god do I feel sorry for you. Custom reporting solutions have often been descriped as a pile of crap. Business Objects, the current owners of
> Crystal Reports (Formerly of Seagate, Crystal Decision, and lord knows what other companies) has long been considered the best of these options. In
> otherwords, they were/are the best piece of crap solution available. Microsoft (insert booing and hissing here) has finally decided to create
> their own reporting solution (SQL Reporting) which from what I've heard is a significant improvement over CR XI (yes, 11 versions and there are
> still limitations on Cut and Paste)

A few points:

1. business objects has a meta-driven reporting platform, and is one of the top three vendors in the OLAP marketplace.

2. it didn't get there by buying crystal reports - crystal reports is the lowest-end product in their suite

3. the microsoft reporting solution is merely another low-end reporting solution, competes with Crystal Reports, but not the rest of the Business Object suite.

4. the business objects BI (business intelligence) platform is very powerful, and is a reasonable tool to use if you are supporting or performing a vast amount of reporting (lets say you're a financial analyst for a large company). otherwise it's often overkill. Note: there are no cheap/free equivilents to this product - the other commercial ones are sold by Cognos and Microstrategy. These are products that are often sold for hundreds of thousands of dollars.

5. describing custom reporting solutions as piles of crap is about as reasonable as describing custom web sites as piles of crap. Some are better than others. The hard thing for many programmers to get their head around is that when you need to analyze a business process in order to get a good idea of what's going on:
    a. you really don't want to have to marshall a few million java objects
    b. you really don't want to have key-value pair tables in your model
    c. you really don't want to think of data as merely persisted objects
    d. you really don't want to have to use an oo database
    e. you are much more likely to drive mysql into the ground due to its lack of parallelism
    f. if you run the reports on a typical transactional data model you'll do millions of joins and the sql will suck
    g. back to f, and the performance will suck, and your server will die
    g. back to f, and the functionality of your analysis will suck due to lack of historical data and lack of integrated data from other systems

Reporting is a discipline with its own best practices, patterns and anti-patterns. Unfortunately, most non-reporting people still think that painting reports via Crystal Reports is a neat idea. Which it hasn't been since around 1995.

Re:Crystal Reports for Java?

[kpharmer]

> Err, I'm afraid I don't follow your analogy there. Now, if there were like 4 available custom web site tool sets, and you had to pay
> $800 per dev seat up front, and a $300 yearly upgrade fee, and those tools all sucked. Then yes, I would agree with your analogy.

Those aren't the only options. What are reports? Well, they can be carefully formatted charts/graphs/tables that print nicely. But that sucks, most people don't care about print any more. What else are they? Tables with pagination, sorting, and navigation along with charts and graphs with similar navigation.

You can easily create very professional, very easy to use online olap-oriented reporting sites using something as simple as PHP & Chart Director. Just focus on a site-map that looks like a genuine OLAP cube, just flatten it out like you'd flatten a globe into multiple 2D maps and then link them together well. And it won't be crappy.

> a) I don't
good

> b) Why not?
Key-value pair tables can't be flattned out. You can't use them as dimensions to filter or sort your queries. Well, not without a ton of sql. They're very handy sometimes for applications, but just don't work for reporting.

d) Why not?
oo databases pretty much died when it became clear that they couldn't handle the massive table scans involved in reporting. At first they brushed this off as completely unnecessary for an application, but eventually people got tired of spending a lot of cash on an OO database, and then spending another lot of cash on a relational database to support reporting.

g2) Entirely dependant on the database design.
a single transactional database seldom has complete history - it's both a performance impact, and complete pain in the butt to maintain historical versions of every row in every table. Setting up foreign keys, writing sql, etc, becomes a nightmare. Further, a single transactional database seldom has much data from other related transactional databases (unless it's an integration-oriented one like a CRM database). But you need history for trending, and you need data from other transactions for more data points.

> You're a data warehouse guy, I get it. And I love being able to use warehouse designs in specific situations, but its not the end
> all be all solution. Warehousing is awesome for reporting, but it is a pain in the ass for data management, security, integrity, and space.

that depends on the implementation: a data warehouse is just a consolidation & distribution point for data. It's great for archiving data efficiently, security usually isn't a problem, and it is an *amazing* tool for finding data quality problems in source systems that you'd never find otherwise.

Now a data mart on the other hand (what a warehouse publishes to) can be a horrific space hog. And is complex to tighten security down on. But there are approaches to these problems as well.

> That's why a lot of data warehouses are based off of summarized OO databases.

hmmm, i've never encountered that. Do you mean oo applications sitting on top of relational databases? If so, I see that all the time. But if you mean an actual oo database, hmm, haven't even run into one in four years now...

> But, we have another application. An internal/consultant developed Commission application. The thing is huge, tracks hundred of
> sale members, service agreements, commissions, splits, charge backs, payouts, etc... And it was designed from the ground
> up with and emphasis on historical data. They can pull up the exact and complete history on every item, sales person, accessory,
> department, office, order, etc. The trade off however is performance. Like you mentioned, even on a beefcake server, it is
> impossible to gather full real time data for reporting. So a series of data warehousing processes are run at night, or after
> changes to a record that alter multiple entities.

Right - I didn't mean to imply that you sh

I'm surprised that it took so long

[brokeninside]

The Eclipse project has been working on business reporting for quite some time. They presently have a rudimentary BIRT module (Business Intelligence and Reporting Tools) that, for its limited breadth and depth, is actually fairly impressive. One of Business Objects' competitors, Actuate, already has a product built on top of Eclipse.

Hopefully, this shift will pan out as a move to better integration of Crystal Reports with web services without having to shell out for Crystal Enterprise. Up through the present, most of Crystal's eggs have been placed in the COM basket so that reporting is best automated through Windows programming. This is great in that you can automatically connect to a database, run a report, export the output and email the export in a few dozen lines of VBScript. But if Business Objects is moving to web services, it will offer a great deal more flexibility as automation will no longer be restricted to Windows.

Eclipse Is The Most Import Software Product

[Anonymous Coward]

I use to think Firefox was the big one, the computing world mover.

Firefox is kid stuff compared to the impact Eclipse is starting to have. My entire company's development has been unified under Eclipse. Developers can seamlessly move from platform to platform. Writing extensions is trivial and only requires a moderate amount of Java experience, which most already have or can have quickly.

Eclipse has suddenly made Linux a first class development platform. Eclipse has turned a huge number of Windows engineers into Linux engineers who write for Linux first and run Linux at home. I can't describe the feeling of freedom and cleanliness that has brought to our company.

Thank god I will never have to touch that piece of garbage Visual Studio again.

Re:Eclipse Is The Most Import Software Product

[G-funk]

Eclipse is a monstrous, horrible piece of software that I hope to never need to use again. IntelliJ forever!

Could this be the vi/emacs argument for the 00's?