I was thinking about learning WF. Now I'm not sure. Would you care to post links to the articles that changed your mind about using it?

I will use it after 4.1 has been released

As requested... here are some of the links...

http://visualstudiomagazine.com/columns/article.aspx?editorialsid=2949

http://visualstudiomagazine.com/listings/list.aspx?id=588

http://sergeluca.spaces.live.com/blog/cns!E8A06D5F2F585013!3233.entry

http://www.dotnetconsult.co.uk/weblog2/PermaLink,guid,707e4cb9-6111-4069-96 b3-596e87f2e262.aspx

http://www.microsoft.com/presspass/events/pdc/docs/NET4OV.doc

http://rickgaribay.net/archive/2008/10/28/wf-4.0-big-changes-ahead.aspx

http://channel9.msdn.com/pdc2008/TL17/

...and the same thing happened with SQL Server DTS 2000 / 2005 and the MapPoint object model and ADO (how many times do we have to re-learn things we used to know how to do)? I'm a strong believer in the MS developer platform and the tools really do get better over time but using us as beta/field testers ain't right. It costs us and our companies real time and money. Now I have Biztalk, WCF, WF, and LINQ on the 'wait and see' list - more will be added I'm sure. If MS still believes that developers are the foundation of its business, it needs to reconsider this strategy. Gates himself talks about all the great companies of yester-year that couldn't see when they started doing things wrong (DEC, Wang, Novell - General Motors?). If MS wants to be arrogant with their competitors (like Oracle or Google) that's fine but its a mistake to alienate the developer community. Sooner or later a 'Toyota' will come along. (Just imagine MS looking for a billion-dollar government bailout. Stranger things have happened)!

Tad,

What do you think of Oslo? See http://msdn.microsoft.com/oslo and http://tinyfinger.blogspot.com/2008/12/foundational-work-on-dsls.html

Has the industry embraced DSL's so far? Is it doing it now, or will it in the foreseeable future? Is there any practicality in V1 of this technology or is it just an amazing exercise of CS that belongs in the research and academic spheres? I wouldn't like to invest an inordinate amount of time in Oslo just to learn that it would have been wiser to wait until V2 or V4 of the technology, as is the case with Entity Framework and WF respectively.

What do you think? Thanks. Fernando

I will have to defer... I am not looking at it seriously for sometime to come.

This is a good unbiased opinion

http://martinfowler.com/bliki/Oslo.html

"I wrote this blog on the topic sometime ago. It appears we cannot blindly trust that releases from Microsoft will be around for more than 2-3 years without becoming obsolete or needing to be completely re-written."

Can't agree with you more. I start considering a technology and then I hear that the technology isn't going to be around or that it sucks.

Wow, what an extreme article… Just because WF is being rewritten does not mean that WF 3.X is obsolete, especially if you look at exposing WF as services. Services that are autonomous should be able to communicate with other services, without concern of which version of WF is running underneath the hood. I am also in the same boat as others who have invested years in building appropriate solutions using WF 3.X, but I do not see this as that big of a deal, more of a shock factor. Please ping me if I can help!

Linq to SQL and WF 3.X are two of the latest technologies from MS I have forbidden my team to bring to the architectural table. EF 1.0 is another one. Given their current track record I am hesitant to look at anything new from them.

My team is responsible for recommending best practices and architectural guidance to a lot of development teams. We cannot in good conscience recommend a technology Microsoft has committed to support part time (Ling to SQL), or one that offers interop as the upgrade path to a new version (WF).

Suggesting services as the solution for integration purposes is a fine legacy wrapping solution. I repeat legacy. I cannot recommend a solution today that is legacy before they even install the library.

Microsoft should get more creative. They should have called the new WF 4.0 libraries Turbo Workflow or something corny like that, and acted like it was a new addition to the .NET Framework that enhances workflow. But they didn't, and it is not. It is a replacement for a mistake.

Do you recommend any off-the-shelf workflow engines to fill the gap until WF4.0 comes out (to be replaced shortly thereafter by the all-new WF5.0, no doubt)?

@ David Keaveny Currently we are advising all our projects to use BizTalk. That is convenient for us because we have a license, but would probably be impractical for most of our projects if we didn't already have it.

I have not investigated any other workflow products.

As one of the articles referenced in your post I can to a degree agree with what you are saying - if you have made a major investment in 3.5 then this sucks. However when WF was written it was written against a bunch of internal requirements from people like the SharePoint and CMS teams. They never envisioned the kind of uses that some people would try to put the library to. To meet this new set of requirements meant a rewrite which hurts people but its better now than when there is wider adoption. The 3.5 runtime will continue to be shipped to support 3.5 workflows and 3.5 activities can be used in 4.0 workflows via the Interop activity. Microsoft generally have a very good track record in backwards compatibility compared with most vendors - For example in my experience Oracle tend to break backwards compatbility with every minor release. Unlike most vendors people build software against every aspect of Microsoft's technologies - trying to get an API right first time every time is impossible. In the Java world you see the carnage of dead open source projects that were at one time or another "the latest thing" - what if you had invested your architecture in one of those? The rewrite is not pretty but people got value out of 3.0/3.5 and they are not left out in the cold with 4.0. And of course BizTalk is only an alternative for certain types of applications - its not going to help you drive the wizard functionality in your WPF application whereas WF can.