According to an article from ComputerWire in theregister.co.uk, WebGain is in the process of selling its TOPLink O-R mapping tool to Oracle and will exit the Java tools market.

For more information, see the original article at http://www.theregister.co.uk/content/53/25579.html.

Chuck McCorvey

After calling a few people about this to verify its authenticity, I didn't get a yes or a no, which would lead me to believe that this is really happening, and the two companies are bound to silence until the deal is finalized (typical of such transactions).

Oracle owning one of the major O/R vendors would significantly propell the attractiveness of its J2EE software stack (neither IBM nor BEA have anything like it), as well as giving it yet another entry point for its salesforce to sell more Oracle9iAS licenses ( Toplink has a huge userbase and continues to be a good seller).

The battle for marketshare got a bit hotter today... :)

Floyd

I wonder if Oracle would be tempted to make TOPLink their default CMP engine. That would definately give them an instant advantage. Will TOPLink developers fear getting locked into TOPLink's engine now it is "owned" buy a vendor (a fear that people had with WebGain)?
Maybe a CocoBase can step upto the plate to show vendor neutrality here.

I really hope TOPLink continues to be a top quality product. It doesn't seem to have moved a lot since WebGain bought it, so maybe Oracle can push it a long better!

Since TOPLink was already providing a noncompliant preview of a JDO interface, I wonder if Oracle will bring it to JDO compliance and be the first of the big relational database vendors to have their own JDO implementation. (The preview was not JDO 1.0 compliant as the spec wasn't finalized when they delivered it.)

--Matthew

To have a major RDBMS/OR Mapper/AppServer vendor provide official support for JDO would be fantastic.

Hi,

    I think this is another indication that Oracle considers Java the best way to interact with their database in preference to PLSQL.

    I don't even try to use Oracle because the version 8 I tried under their OTN trial program did not support auto-incrementing integer primary keys, without setting up sequences and triggers, which was a huge pain. My attempt to use the Migration Workbench from MS SQL Server was a disaster as a result. Does anyone know if Oracle has fixed this or has plans to?

    Thanks much for the topic. I suppose this is as good a time as any to thank all the people at Symantec and WebGain for their VisualCafe IDE. It is what I started with, and enjoyed and learned from tremendously. That was late 1996. In the fast-moving world of development tools, it was overcome, but it will always be fondly remembered as a key contributor to the naiscent Java revolution.

Tom Schaefer
Senior Technical Expert
Tecolote Research, Inc.

I think this is another indication that WebGain is screwed

I wonder how this affects BEA. They seemed to have some significant investment in WebGain.

Not sure about your comment on setting up Sequences for auto-generation of primary keys. I've always thought it was pretty easy and rock solid. You're guaranteed that no matter how high your trnsaction rates, the sequence will always generate keys correctly.

Much better than other solutions and easy to set up (I thought at least).

As far as I know, the Oracle Migration Workbench does auto-generate sequences and triggers to simulate autoincrementing IDs. I remember reading in the latest Oracle Magazine that someone migrating from Sybase used this capability.
You may want to check on the Oracle Technology Network.

Hiya Dion :-),
First of all, sorry for that bad result of England in the world cup:-). Be confident that if Oracle gets TopLink (and i am not saying that it will) it will be pushed a way long better...BC4J is a good proof of how far a J2EE framework can go in Oracle...however I still don´t see very clear the JDO integration though...I bet you have been following the SQL-JDO battle...tough issue. We are going to need more good samples from ya.:-)
Cheers!

Dion,

Thanks for your note, and yes THOUGHT Inc. and the CocoBase
product are committed to vendor neutrality and to providing
the best possible O/R for Java while integrating with as
many tools and platforms that are possible.

CocoBase even works with the Oracle App server and DB, so customers can have their cake and eat it too :)

To show our commitment to this principal, the latest
version of CocoBase includes vendor neutral Facade
interfaces to our O/R layer as source code to ensure that
customers can use and even wrapper the CocoBase runtime
with a completely neutral and even customer extensible
layer! We provide 2 implementations currently, one that's
standalone, and the second one that uses Session beans.
With this approach a developer can write a single Object
model and deploy an app locally or on a J2EE server simply
by connecting differently!

Have fun and happy java programming!

Ward Mullins
CTO
THOUGHT Inc.
http://www.thoughtinc.com

Talking aboout competition and moves in the market, I wonder how does it come that this piece of news goes un-noticed:

"HEWLETT-PACKARD IS ABOUT to sell off its middleware assets and exit the business, looking instead to partnerships to meet its customers' middleware needs, HP executives said Tuesday."
Read more

It seems that they are looking for a buyer and they won't comment on the application server directly: "Chief Executive Officer Carly Fiorina said HP won't provide concrete details until customer transition plans are in place. "

Interesting times.
 

"auto-incrementing integer primary keys, without setting up sequences and triggers, which was a huge pain. My attempt to use the Migration Workbench from MS SQL Server was a disaster as a result. Does anyone know if Oracle has fixed this or has plans to? "

This is by design; identity columns have traditionally had a lot of problems when using them in transactions (The @@IDENTITY variable in TSQL was tricky to use correctly, for example). I wouldn't class this as a "major pain", its more of a design decision to get used to.

As for this transaction, I think it's great news. WebGain really didn't push TOPLink for the first while they owned it, only recently did they start. I think they did a decent service to the product in improving its GUI, but there's still a lot of work to do.

From a competitive angle it's interesting because WebLogic 7 already imbeds a lot of TopLink's ideas into its CMP engine.... I would think that Oracle would benefit more if they embedded TL directly in the database instead of just supplying it with 9iAS. Or they could just keep it as a separate product.

<Stu>
I would think that Oracle would benefit more if they embedded TL directly in the database instead of just supplying it with 9iAS. Or they could just keep it as a separate product.
</Stu>

If they did this wouldn't it just be GemStone? ;) Ah, what a perfect world it would be if this happened...

In search of nerdvana,
Greg Peres

What ever Oracle does, I just hope they don't follow Webgain's pricing model for TL. Didn't anyone ever teach them economics, 3 customers @ a gazillion dollars or 1000 customers @ something respectful and they still make more money.

What a great product, but it was just priced right out of most shops.

Chuck

<Chuck>
What ever Oracle does, I just hope they don't follow Webgain's pricing model for TL. Didn't anyone ever teach them economics, 3 customers @ a gazillion dollars or 1000 customers @ something respectful and they still make more money.
</Chuck>

Tend to agree but somewhat disagree with you. I agree that TopLink is in an "elite product" price range but - that is all for a good reason. The product does very complex things that most can't do. Mind you most consumers don't need those features so can't understand why they have to pay the price.

But you do bring up a good point. They could have a bigger market share if they put a better price on it. But then they would need a bigger support team, possibly less innovation, gamma et all...

Law of diminishing returns.

"When increasing amounts of one factor of production are employed in production along with a fixed amount of some other production factor, after some point, the resulting increases in output of product become smaller and smaller."

Listen to more Fantomas,
Greg Peres

Just remember Greg that we don't know where the point of maximum profit is for their product, so it's not completely as cut-and-dry as the law of diminishing returns says. I would hope they have figured this out though...

n.t

<nile>
Just remember Greg that we don't know where the point of maximum profit is for their product, so it's not completely as cut-and-dry as the law of diminishing returns says. I would hope they have figured this out though...
</there>

Good point but the difficulty for many software (computer) companies is that once they cross the point of maximum profit they enter a huge risk zone.

That is why you see these kinds of prices on products. In order to avoid crossing the maximum point they over price their products to make it attainable to a mass market. It is an avoidance of a company ending mistake. If your out of their target market you complain it is too expensive. :D

I guess my point is: moreCustomers != moreProfits

It is all about balance and in the software industry and it is very hard to find that balance. The market also moves the point of maximum profit on a day to day basis making it even harder to estimate. So companies keep a safer more secure spot at the expense of missing out on a few potential buys at a lower price point.

It’s about good business not bad prices.

The plot thickens every day,
Greg Peres

Correction:

<Greg>
In order to avoid crossing the maximum point they over price their products to make it attainable to a mass market.
</Greg>

In order to avoid crossing the maximum point they over price their products to make it unattainable to the mass market.

Things always look different in the text box when you’re writing them.

The pieces of the puzzle keep slipping away,
Greg Peres

Greg, I do agree that it is a great product. However, it's not a luxury car. Software organizations can't afford to have discriminating tastes. They won’t put off that luxury purchase until a later date. The project won’t allow that. They simply will walk down the virtual street and buy Cocobase, or roll their own. Even during the down market, WebGain continued to charge outrageous prices, even though products like Cocobase were gaining market share.

I worked for a .com during this time and they wanted us to forecast a year out the sales of our J2EE product and then charge us if we didn’t hit the numbers. It was insane.

Chuck

"I worked for a .com during this time and they wanted us to forecast a year out the sales of our J2EE product and then charge us if we didn’t hit the numbers. It was insane. "

I'd say that's good business. They're not out to give something away for free.

Furthermore, as much as Cocobase is a decent tool, there are some things TOPLink does that it doesn't do. And TOPLink arguably does a lot of thing better than Cocobase traditionally has. Now you may argue that "business people don't care and will buy the cheaper product". I will suggest that fine -- but wtf are business people making technical decisions? It happens, sure, but it's the epitomy of disfunction to suggest that a go-kart is comparable to a Covertte.

Stu,

Your claim about CocoBase lacking features against Toplink
seems a little off the mark. Generally the opposite is
true. After all CocoBase supports about 30 App server
versions currently, which is about 15 times more than
what Toplink supports... CocoBase integrates with 3
different IDEs, virtually every modeling tool available.
It provides the only portable and extensible CMP solution
of any O/R tool. It provides a 100% extensible plugin
layer including caching. I think may have gotten that
statement backwards...

Oh and lets not forget that THOUGHT Inc. is profitable
and doesn't need Oracle to bail us out from an unstable
and unprofitable business model :)

Are you sure you evaluated CocoBase, or did you just take
the Toplink salesperson's word about these supposed missing
features?

I'm curious, can you name a feature that's in Toplink that
isn't in CocoBase or was this just a general perception
that you had?

Just my $.02
Ward Mullins
CTO
THOUGHT Inc.
http://www.thoughtinc.com

<Ward>
Oh and lets not forget that THOUGHT Inc. is profitable
and doesn't need Oracle to bail us out from an unstable
and unprofitable business model :)

Are you sure you evaluated CocoBase, or did you just take
the Toplink salesperson's word about these supposed missing
features?
</Ward>

As the CTO of the company I would expect a much better responce from you. You might want to follow Damodar Periwal's example from Software Tree and not flame potential customers.

My 2 cents worth,
Greg Peres

<Greg>
As the CTO of the company I would expect a much better responce from you. You might want to follow Damodar Periwal's example from Software Tree and not flame potential customers.
</Greg>

Greg, I would hardly call it a flame, I simply disagreed
with Stu, or do you think disagreement is always a flame?
Some folks do...

Unlike Periwel, Stu made claims about CocoBase that are
false. Perhaps you thing this is professional, but I do
not... We just have different ideas of what's acceptable
I guess...

I'm still waiting on the claims to be explained, or do you
think that's unreasonable? I personally believe if you
make those sorts of claims, you should be able to back them
up - don't you think that's reasonable?

Just my $.02

Ward Mullins
CTO
THOUGHT Inc.
http://www.thoughtinc.com/

oh no, please not that discussion again. How many times before have I seen this pattern - and why is it always Wards name that appears prominently in that context?

Let me ask one thing about this statement, however:

<Ward>
Oh and lets not forget that THOUGHT Inc. is profitable
and doesn't need Oracle to bail us out from an unstable
and unprofitable business model :)
</Ward>

I have seen too many projects using TopLink (and paying their license fees) to be able to believe they run a "unstable and unprofitable business model". Can you substantiate your claims a little more?

Peace,
Christian

<Christian>
oh no, please not that discussion again. How many times before have I seen this pattern - and why is it always Wards name that appears prominently in that context?
</Christian>

Maybe it's because a hand full of folks like to make
unsubstantiated claims on theserverside, and I personally
find that unprofessional and am willing to say so. You
may have different values, but I find it unethical and I'm
willing to stand up and say so. You may think it's the job
of vendors to take it on the chin from engineers who have
opinions that are patently false, but I don't... But then
free speach is still a right and a beautiful thing isn't
it? You have a right to state your opinion and I have a
right to refute it - don't I? Or would you like to take
that away as well?

<Christian>
I have seen too many projects using TopLink (and paying their license fees) to be able to believe they run a "unstable and unprofitable business model". Can you substantiate your claims a little more?
</Christian>

You might want to read the posting before criticizing it.
I was referring to Webgain being unstable & unprofitable as
is stated by their loss of $200M in the last 2 years.
Basically Toplink was bought by Webgain because the founders
ran out of Money to support it, and the same thing happened
with Webgain. To me that seems like an unprofitable
business model when the product has to continually be passed
around like a hot potato - don't you?

There's something to be said for a company that actually
runs a real business, watches costs and the bottom line
and is profitable (that would be THOUGHT Inc. as I'm sure
you've gathered by now)...

And as far as I'm concerned, I'm done on this topic. The
bating as fun guys, but I have a company to help run,
especially in light of the fact that our major competitor
just crashed and burned in a marvelous blue flame :)

Have fun - and happy java coding.

Just my $.02

Ward Mullins
THOUGHT Inc.
http://www.thoughtinc.com/

<Ward>
Greg, I would hardly call it a flame, I simply disagreed
with Stu, or do you think disagreement is always a flame?
Some folks do...
</Ward>

You implied that Stuart was manipulated by a WebGain sales representative and that was the basis of his comment. Quite a sweeping assumption.

I don't always feel disagreements are a flame. I just feel disagreements tied to insults are. You may not have intended to insult Stuart directly but it came across that way to me. I apologize if I miss interpreted what you wrote.

But this is not the best place to discuss the topic of what a flame is...

I do expect Stuart to write an excellent response. Typically his Wiki Web responses are typically very good and quite educational.

Can't we all just get along?
Greg Peres

<Greg>
You implied that Stuart was manipulated by a WebGain sales representative and that was the basis of his comment. Quite a sweeping assumption.
</Greg>

Actually I didn't assume, I asked if that was what happened
or if he had actually done an evaluation... Before I
responded I checked our request logs, and we don't show
that Stu ever evaluated CocoBase, so I think it would have
been a safe assumption even if I had made it - don't you?
But in reality I asked where he obtained his knowledge, and
generally customers who have this 'opinion', do obtain it
from Webgain, and not from evaluating CocoBase. And if they
did evaluate, they simply did so cursory and didn't ask
questions or read the documents thoroughly. Most problems
come from developers who make assumptions and don't read
documentation, and instead rely on quite inaccurate
'competitive' documents that are written to mislead, or are
severly outdated or filled with bad assumptions...

So in fact if you read it as an assumption on my part, that
would seem strange as I asked where he obtained his
information - an assumption wouldn't have done that would
it?

And I disagree that this isn't the place to discuss what a
flame is, as there seems to be a lot of disagreement from
those who post here on what it actually is. Thank you for
apologizing for the misunderstanding, and I hope you
understand that I believe that if you're going to criticize
a product, you need your facts in line - don't you?

Just my $.02

Ward Mullins
CTO
THOUGHT Inc.
http://www.thoughtinc.com

"Before I
responded I checked our request logs, and we don't show
that Stu ever evaluated CocoBase, so I think it would have
been a safe assumption even if I had made it - don't you?"

I've evaluated CocoBase at least 4 times since 1999, at 2 different companies. I entered my log information each time.... I have no idea why you can't find it.

<Ward>
But in reality I asked where he obtained his knowledge, and
generally customers who have this 'opinion', do obtain it
from Webgain, and not from evaluating CocoBase. And if they
did evaluate, they simply did so cursory and didn't ask
questions or read the documents thoroughly.
</Ward>

I understand your argument. As a vendor you are often the victim of your competitions marketing department. I know how misleading it can be for many developers.

Also, people get so zealous about the products they in this industry. Especially the people on Java Zealots. ;) (Sorry Floyd had to do it. :D )

I too have the same issues with many of the people commenting on many of the subjects here. Often I give up too often and not bother to contribute.

<Ward>
Thank you for apologizing for the misunderstanding, and I hope you understand that I believe that if you're going to criticize a product, you need your facts in line - don't you?
</Ward>

I do agree. I think Stuart has responded with an excellent list of questions.

All I wanted was a Pepsi... Just one Pepsi!
Greg Peres

<Gregory>
I do agree. I think Stuart has responded with an excellent list of questions.
</Gregory>

I agree Gregory, thanks for your response, and I REALLY
appreciate folks like Stu posting their issues and concerns
in a way that makes it so easy to respond with facts and
not just finger pointing. It's so much more productive
for everyone involved, and I would encourage engineers to
do the same as a rule. Another thing that works for me
is the phrase 'from what I understand', or 'correct me if
I'm wrong'. That REALLY works for me :)

This thread has certainly improved from the standard flame
stuff that I generally see from Christian and some of
the other folks who seem to like to pick a good fight more
than actually getting to the truth. Growing up in a big
family, I'm not afraid of a good fight, but I don't go
looking for it... Especially when the 'can't we all just
get along' approach seems so much better for everyone...

I'm never sure what axe some folks have to grind, but it's
often disappointing to see displayed for all...

Hip Hip Horray for you Stu!

Just my $.02

Ward Mullins
CTO
THOUGHT Inc.
http://www.thoughtinc.com

Ward, on an unrelated note - please don't press enter at the end of each line in the 'comment on a news article form'. Thats why your posts don't stretch out like everyone elses in this thread do.

You only need to press enter between paragraphs.

Floyd

Hey, I'm just here to hash out what the best solution is. I've had good success with TOPLink and don't agree when people feel its price in unjustified. Obviously Ward has a different & valuable perspective in that he's dealt with many who are frustrated with TOPLink, that adds balance to the debate.

I'm just stating an opinion & not afraid to debate its validity on the merits. :)

Stu

Stu & Ward,

Thank you for a very informative and lively argument. This is a perfect example of how we can learn from disagreements. Kudos to both of you for not letting this get out of hands.
Since, I am fairly new to the topic of O/R mapping and their integration with modern app servers, can either (or both) of you recommend a good source (book(s)/article(s)) where I can learn more?

Thanks,

- Adnan Rafiq

<Adnan>
Thank you for a very informative and lively argument. This is a perfect example of how we can learn from disagreements. Kudos to both of you for not letting this get out of hands.
</Adnan>

Glad you like it... It helps us when folks are like Stu who have done their homework... And it makes sure we're doing our job product wise to satisfy product needs - the reason we're still in business after almost 10 years when most java software companies have gone the way of Webgain!

<Adnan>
Since, I am fairly new to the topic of O/R mapping and their integration with modern app servers, can either (or both) of you recommend a good source (book(s)/article(s)) where I can learn more?
</Adnan>

Sadly there are none that I like. There's so much academic stuff written with so little actual understanding of the real issues, that I'm usually left horribly disappointed by what I read. There is so much confusion over 'what is O/R mapping', and so many large companies have 'claimed' to support it when they don't seem to even know what it is that it leaves the literature and body of knowledge very incomplete or simply wrong...

Like I personally don't consider the embedded SQL generation of java code 'O/R Mapping'. Many companies sell products that only provide this functionality and use the term O/R mapping - which I find misleading and deceptive. That's why we've begun to use the trademarked term 'Dynamic O/R Mapping' to distinguish our product from that class of technology. Hardcoded and embedded SQL systems tend to have really crappy longterm Return on Investments, please have a look at the ROI of CocoBase (it's on our website), it's quite astonishing in contrast.

I have read some good articles in the past, and in particular Scott Ambler seems to have a great grasp on the subject area. Most folks who write on the subject however seem like they're all re-writing the same midterm paper from crib notes and it's simply a bunch of bunk...

Are there particular questions you are interested in? Maybe we can get Floyd to start a new thread and get some more indepth issues pushed out into the web... I think there's enough interest to warrant this...

But thanks for the interest in both CocoBase and O/R mapping and your kind comments - always glad to help! I agree that it is much more enjoyable to participate in this kind of discussion, and probably a lot more 'rewarding', although perhaps less enjoyable for many of the readers :)

Ward Mullins
CTO
THOUGHT Inc.
http://www.thoughtinc.com

Hi,

after another interessting read from people like stu, ward and others, i would step back to the original thread and
want to note that webgain canceled ALL "TopLink free one-day workshops" in germany.

This is IMO another point for the truth of the original thread.

Marc Logemann

As for 'where can I read-up on O/R Mapping'-- Stu is right to suggest Ambler (which is generally a safe bet)...

Here's a link:
http://www.ambysoft.com/mappingObjects.html

I'd also suggest ODMG's stale but useful whitepapers link:
http://www.odmg.org/library/whitepapers.htm
(see the section called 'Object Relational Mapping'.)

Stephen Henderson

Ambler is good. Also look at some of Apple's documentation of WebObjects, I've always found they had the easiest to grok framework & toolset.

Ward,

First off let me say that I am in no way trying to defame your product. It is a solid product and clearly improving at a faster rate than TOPLink is. I just happen to think that TOPLink is the better product.

I find your suggestion that I'm really just spouting a salespersons' word a shallow and ignorant cheap shot, and indicative of a poorly thought out counterpoint. Criticisms aren't flames, but claiming that I know nothing when you don't know me is definitely a flame. I'd suggest talking to several members of this site, including Floyd (whom I've worked with), about my expertise in this area. In no way am I spouting anything that is "patently wrong", and I challenge you to point it out. Oh wait, "you're done with this topic", ducking out before I can respond. I'm supposed to be the unprofessional one?

On the other hand, the Go-Kart -> Corvette comparison was a bit harsh, and mainly for dramatic effect, and probably Cocobase is at least comparable to a VW Turbo Beetle.

Having said that, I'd like to suggest a re-read of my post. You'll notice that my point was that TOPLink solved problems better than the way Cocobase *traditionally* has. Your transparent dynamic persistence is a new feature that has always been in TOPLink, and I’m glad that you now have it. I’m by no means an expert on Cocobase, so perhaps you can help me out here and answer some of my concerns. Perhaps you’ll convert me. :-)

There are still general "taste" issues involved between the two products. I prefer the UnitOfWork API over the Navigator / Links API. Naturally the bi-directional maintenance of links in CocoBase is nice, but EJB 2 CMP Toplink does this as well. It is still unclear to me whether Cocobase covers many of the features I find important in TOPLink. For example, I’m not sure if Coco handles Oracle 8i's object/relational constructs, such as arrays, structs, and nested tables. Also, mapping of inheritance appears to lack pre-defined strategies in the admin tool (though I’m sure one could do it anyway).

 Another concern is polymorphism: if I have a 1:M or M:M relationship between heterogenous objects (all in the same inheritance hierarchy, identified through a type code in one of the columns), can CocoBase ensure that it instantiates the proper objects when resolving that relationship? I.e. I have a relationship to an order book with different kinds of Order objects (FullSalesOrder, DirectSalesOrder, InternetOrder, etc). Would I get the right instance in my collection? TOPLink can do this.

How about derived columns & transformations? Would I have to throw out transparent dynamic persistence & CBProxyM and just implement CBProp, like old-style Cocobase? (which is, as I mentioned above, my biggest traditional peeve with CocoBase). Certainly I can intercept the incoming data on setPropObjectData() and then transform it, but TOPLink allows me to create hooks & transformation methods on a per-field basis.

Can I have relationships in Cocobase as Maps? Do they have to be Collections?

Does your object querying support a distributed cursored collection? I.e. if I call a remote EJB finder, can you stream the objects across the 3rd tier, or do you have to query ALL the rows, populate them into objects, and then pass the WHOLE collection across RMI/IIOP? TOPLink allows me to get them piecemeal with their CursoredEnumerator / CursoredCollection.

How about your dynamic query API? I don’t see the chapter yet in the version 4 programmer’s guide. Can it do correlated sub-queries & map the results to objects? I assume I could just write plain old SQL to do this and Cocobase would figure a way to map those results to a partial or full object of my choosing, as long as I specify the map. Is this right?

Do you support full EJB QL? Even MEMBER OF? TopLink 4.5 doesn’t fully support it yet, this would be a good thing. Not sure about WebLogic 7’s CMP engine ,it might.

I have no interest in WebGain as a company, and my only affiliation to TOPLink is that it's written by Canadians, and I'm a Canadian. <grin>. I think WebGain was a poorly run company, but that observation has very little to do with the quality of their product (which was due to The Object People) -- mainly due to the lack of quality in WebGain Studio and Visual Cafe (just an opinion, again).

Cheers,
Stu

Stu,

It sounds like you've definitely looked at CocoBase, too bad
you didn't speak to our support staff as they could have
corrected & explained many of your misconceptions. And may
I ask how you have a copy of CocoBase without having
requested it officially from what I was able to see? Did
you use an alias id? And I do thank you for putting up your
concerns and assessment, as it gives me something solid to
respond to instead of the all too common thread on this site
of 'it doesn't do everything', that's entirely too vague
and filled with just opinion...

Now as for your technical questions, lets take them one at
a time:

First of all you many of the items you list are very clearly
not 'issues' or 'features missing', but design choices. In
particular saying CocoBase has a particular API that you
like, hardly qualifies as a feature when we have another way
of accomplishing the same task...

Your statement said that CocoBase doesn't do things that
Toplink does do, and after reading this assessment I can
tell you that this still isn't true from the list that you
have presented here. It seems you simply don't like 'how'
CocoBase does it - but that's not the same thing, and I
have a problem with this positioning if you feel that way.
So lets break out your items and see why CocoBase still
does more, and costs less.

<Unit of Work vs. Navigator Links> Actually you might like
the Toplink approach better, but CocoBase has more features
because you can use multiple links with the same concrete
object model, so technically CocoBase is superior in this
feature set. Oh and CocoBase has been delivering the
Transparent persistence for years (perhaps not as well
documented as it should have been), and it isn't a recent
technology. The CocoProxyM wrapper class has been around
for about 4 years, as has CocoNavigate (the predecessor)
to the Navigator and CBFacade APIs. If you like the Unit
of Work APIs better in Toplink, you should look at the
CocoBase CBFacade APIs. They simply wrapper the CocoBase
facilities in a simple API that you'll probably love. The
Navigator APIs are for low level manipulation, and the
Wrapper/Facade APIs are meant for those folk like yourselves
that like a higher level API... But unlike Toplink, we
offer both levels of control...

<object relational constructs> CocoBase fully supports
JDBC 2.0 types including these you've mentioned.

<mapping of inheritance> Please look at the users guide,
and you'll see this fully supported in the admin tool.

<polymorphism> Please look at the users guide and in
particular search for inheritance - fully supported.
Including in the 1-m & m-m relationships. And the
descrimitor facility works out of the box with a variety of
options, and is even user extensible! There's even a
wizard in the tools bar of CocoAdmin to manipulate the
descriminator, and the default implementation is included
as source code that does the actual instantiations...

<Derived columns> CocoBase proxy handling is extensible and
shipped as source code - you can implement the same
features, and in fact you have more control with CocoBase
as you can route through using raw fields or even alternate
accessor methods. Once again CocoBase provides a richer
solution than Toplink for doing this. In fact we provide 3
separate hooks that can all be used for this task. The
wrapper Proxy class, the custom factory mechanism, and even
the Plugin API which can be used for the same task. It
sounds like your issue with us is that we give you too much
control over this - hardly a missing feature though...

<Can I have relationships in Cocobase as Maps?> Yes, and
in fact the facility is extensible with CocoBase and you
don't need those unnecessary accessor methods the way you do
with Toplink in your root objects. You can even specify
what 'key' methods are used, etc...

<distributed cursored collection> Yes, CocoBase can support
this. Most app servers cannot however because of thread
and connection pool issues. In those cases you can use
CocoBase and 'simulate' the behavior by 'jumping' to the
correct location for each block and reading from that point.

<Dynamic Querying> CocoBase has apis that even let you
create maps on the fly, you can do basically anything you
want, and we ship a querying implementation as source
code.

<EJBQL> This is the only thing you could call 'missing',
but since it's simply an alternate querying facility, and
there are other options, it's not a 'missing' feature
in the sense that there isn't another way of accomplishing
it...

<Stu>
I have no interest in WebGain as a company, and my only affiliation to TOPLink is that it's written by Canadians, and I'm a Canadian. <grin>. I think WebGain was a poorly run company, but that observation has very little to do with the quality of their product (which was due to The Object People) -- mainly due to the lack of quality in WebGain Studio and Visual Cafe (just an opinion, again).
</Stu>

That's good to know, but considering how many customers
we've had to migrate off of Toplink because of issues, I
have a little different opinion... But I do feel that
the biggest problem of most businesses is the 'business',
and not just the technology and in general I would agree
with you on the most important factor here for product
success and failure...

CocoBase is a much larger product in terms of features,
integrations and extension APIs than Toplink is, so that
sometimes makes it difficult to fully understand, exploit
or document all of the features that it does. Most of
the documentation focuses on the common 80% user, but that
doesn't mean that CocoBase doesn't implement the feature
that you need. Please just ask our support guys if you
can't find something, there's probably an engineering
document hanging out just waiting to be emailed :)

Now please have a closer look at CocoBase when you're ready
to use an Enterprise O/R tool. I can't see anything in
your list that we don't do, so we're probably the correct
choice for your company.

Have fun and happy java coding!

Just my $.02

Ward Mullins
CTO
THOUGHT Inc.
http://www.thoughtinc.com/

Hi Stu,

Just wanted to brief you on how FrontierSuite addresses the questions you have posted for OR tools, compared to TopLink

FrontierSuite provides all the features what TopLink provides in a much simpler and elegant way for the developer to deal with.

<Unit of work>
In FrontierSuite, persistence objects are always within transactional boundaries. Whether you are using FrontierSuite in a managed or non-managed environment, JTA is the unit of work controller in FrontierSuite.

Object by reachability is a core feture in FrontierSuite, and the persistence manager deals with state changes transparently when the transaction completes. The developer has less work to do, to deal with keeping track of changed objects.

In FrontierSuite for JDO, user can have non transactional boundaries for persistence objects which does not need to follow transactional boundaries.
</Unit of work>

<Inheritance and Polymorphism>
One of the mismatch in object to relational mapping is inheritance which is addressed in FrontierSuite through Vertical Mapping, Horizontal Mapping and Collaped Mapping options.

When navigating relationships (1:M, M:M), FrontierSuite returns the hetrogeneous hierarchy objects in the relationships without developer doing any extra work.
</Inheritance and Polymorphism>

<derived columns>
Through code generation and customization life cycle methods can be used to customize the data attributes sent to and/or retrived from the database.
</derived columns>

<dynamic query>
Query management is very powerful and very dynamic in FrontierSuite. It supports OQL, EJB-QL and JDO-QL
</dynamic query>

<cursored collection>
In a managed environment, FrontierSuite does not provide cursored collection, as transaction management is under the control of the container. But in a non-managed environment yes, you can (long lived transactions)
</cursored collection>

<real world scenarios>
FrontierSuite has four development environment to suite the application development. 1) Forward engg 2)Reverse engg from database. 3)Reverse engg java classes 4)Bridging existing java classes and existing relational schemas
</real world scenarios>

<choice of technology>
EJB2.0 or JDO1.0 depending upon your requirements.
</choice of technology>

Hope you have time to have a look at FrontierSuite
(www.objectfrontier.com)

Dominic Savio

Wonder how the O/R purchase by Oracle will play out eventually as it looks like Oracle is making a play for the HP's Bluestone ( aka Netaction ) portfolio :

http://www.infoworld.com/articles/hn/xml/02/06/06/020606hnoraclebuy.xml

" .. ORACLE IS IN discussions with Hewlett-Packard to buy its middleware assets, a move that could provide a much-needed boost to Oracle's application server business, industry sources said.. "

Wonder what this whole episode will have on their existing oc4j based portfolio ? Good thing my current project has nothing invested from Oracle'e end .

As I am currently working on a project that is nearing completion that was ported from WebLogic to Oracle OC4J, this is a risk that we never hoped to consider. My experience with Bluestone has been mixed.

<Chuck>
However, it's not a luxury car. Software organizations can't afford to have discriminating tastes.
</Chuck>

Well, if this were true I would be typing this email on my 390 or Tandem but instead I am typing it on a PC. :(

Chuck, if this news article about Oracle purchasing TopLink is true then you can expect the price to come down. Oracle has a larger support organization and more products that they make profits on to support a lower price. So in the end you may just get what you asked for. But don't expect it to be free. Oracle isn't known for their "cheap prices" in the software industry. ;)

Stuart - excellent point. :)

It's in your face but you can't grab it,
Greg Peres

There are other good alternatives. Check out JDX from Software Tree. JDX is a powerful, flexible and robust OR-Mapping product. JDX employs a non-intrusive dynamic architecture and supports complex object-modeling including class-hierarchies.

Many top-tier companies world-wide have benefited and are enjoying the productivity and performance gains provided by JDX. And you don't have to break your back to license or use the light weight persistence engine of JDX.

Damodar Periwal
Software Tree, Inc.

" I think they did a decent service to the product in improving its GUI, but there's still a lot of work to do."

In what way did they improve the UI. They went from a stable mapping workbench to a buggy one just because the original was written in Visual Age Smalltalk and not in Java. The old mapping workbench was far superior.

Well,

This is good news but for those interested in JDO/Object persistence it is not a big win. In fact I wouldn't be surprised if the JDO support within TopLink disappeared.

Oracle would have to "eat a lot of crow" if they moved forward in supporting JDO or any other object persistence model. Sorry to burst anyone's object persistence bubble but don't expect RDBMS vendors to ever be excited about JDO. It would be like Microsoft admitting they should have stuck with Java. (Well maybe not as dramatic.) There is a lot of negative history in the RDBMS vrs OODMS world.

However I applaud Oracle with both hands for this move. It is a necessary evil for them and it is good that they will be taking over the product. WebGain never realized the products potential. This could be one of the critical tools that could push 9iAS on to the top of the stack.

I have to admit... I would have never really considered Oracle but... This really caught my attention. :)

Cake is good,
Greg Peres

Some day you have to show me the other way to applaud ;-) (no offense, just could not resist).

<Mike>
Some day you have to show me the other way to applaud ;-) (no offense, just could not resist).
</Mike>

It is better than the sound of One Hand Clapping.

Listen to Black Sabbath,
Greg Peres

We have been using ObjectFrontier tools with great results and ObjectFrontier is a also a very good player in the OR mapping and persistence space.
ObjectFrontier offers a very good OR mapping product and their product, FrontierSuite, which we have been using for long time, provides some great features and nominated for readers choice awards. So the closing down of Weblink is nothing very serious.

Yugandhar

CNET is reporting tonight that Oracle has announced that it has purchased TopLink, and will give it away free on OTN.

See http://news.com.com/2100-1001-936185.html?tag=fd_top for the details.

Alan

Yes. How interesting is that considering that BEA and Warburg Pincus (funded BEA's mezzanine round) own the company.

Wonder who's get the better end of this deal?