Wagnerblog

Seriously folks. I downloaded the “Next Generation Data Access” Whitepaper from Microsoft over the weekend just to see what’s potentially coming in ADO.NET 3.0. You know MS has to mess with its data access technology every couple of years –  right? It’s part of the corporate charter somewhere. “Thou shalt remodel thy data access stucture as often as possible”.  After all, the 6 million developers out there need something to do. OK enough with the sarcasm.

Upon browsing the white paper several key points immediatly jumped out:

• The new technology contains an Object-Relational Data Mapper that uses XML mapping files
• LINQ creates dynamic SQL – let me repeat that – LINQ creates dynamic SQL
• According to the References quoted by the white paper, LINQ draws on Dr. Chen’s Entity Relationship Model

These are all very very noteworthy developments. First and foremost it should be noted that most of the technology espoused by LINQ has been perfected in my favorite ORM tool for several years already, and users of LLBLGen are well aware of Chen’s Entity Model. It is the very basis of LLBLGen. So if you are interested in working with a tool that has already gone through 2 versions of releases and has crossed several of the mountains that I’m certain the ADO team has yet to scale, check out LLBLGen.

Secondly, again this is very interesting – dynamic SQL as a “first class citizen” of the data access universe at MS. This is a radical departure to the mantra of stored procedures preached by various evangelists and proponents of SQL Server. A couple of years ago we even had a few rounds of blog comment discussions sparked by a post from the ASP.NET team. Somehow it seemed news to a lot of folks – even MS employees – that dynamic SQL executes as fast as procs.So having MS put out an engine that relies on dynamic SQL is big news.

LINQ uses XML mapping files – what a bummer! I think that’s pandering greatly to Hibernate, IBatis and who knows what else. It does make me suspect that perhaps many enterprise shops have started to reach for those open source efforts, and in order not to become left behind someone at MS decided that using XML mapping files is a good thing. Well I’m here to tell you it is not. Mind you this is not an original thought of mine. Wish it were. Rather it comes from my colleague and friend Dan Madoni who has sloughed through one of the larger J2EE implementations in the country – using Hibernate. And the way Dan likes to explain it, XML mapping code is essentially the same as moving logic out of your code where it can be traced and debugged into an XML file where it is hard to read, convoluted and next to impossible to debug. I never had to worry about that because I’ve had all the shiny goodness of LINQ, Hibernate and IBatis without the gray hair resulting from the XML files by using LLBLGen.

OK so I do sound like a broken record when it comes to that tool. But stop and think for a second here – how many people do you know who in the past 5 years have gone out to create a .NET product that completely flew in the face of the prevailing Microsoft theories espoused by thousands of Regional Directors, Evangelists, Product Managers and every Tom Dick and Harry wanting to get their name on all the ASP.NET sites out there. Frans Bouma has been beating the OO drum and Dr.Chen’s model for just about that length of time. He has compeletely gone against the “theory du jour” and as far as I know had to fight one of the bigger uphill battles that I’ve seen a single developer Micro-ISV fight.  That’s courage of conviction if I have ever seen it.

And the big news with LINQ is not that Microsoft Research developed LINQ and the product team is incorporating its technology. Not at all. The big news around the entire LINQ story for me is how similar to LLBLGen it is and how much that vindicates Frans and all the detractors he had to face over the years. Remember until very recently MS proclaimed the DataSet as king. It is not. Many people in the industry have said so (more recently in the NET 2.0 timeframe than during the release of 1.1). But you know what, for my money MS should save themeselves a ton of dough and just buy LLBLGen. It is so much more mature than LINQ it would make one heck of a value proposition.

UPDATE: While I contrast so much of what I read against the tool I know and love, Oren Eini actually parsed the document and made several very distinct comments that go into greater detail about the potential hazzards and pitfalls of using LINQ. I especially liked “  Oh joy! Three ORM frameworks. Linq to SQL, Linq to DataSet, Linq to Entities.” (via Mike Gunderloy)