The BPF stuff is a great case in point. Privately, they’ve been admitting to big customers for *years* that the "good stuff" in BPF is implemented in Javascript, and that it’s of limited value if you can’t use it out of the box. If your needs include case management complexity, but don’t happen to precisely fit the BPF implementation, you have two options: implement custom requirements in Javascript at the browser level, or do without. I’ve seen both, and neither is pretty. The obvious solution to this is to port the BPF code to Java and make it a first class part of the back end, with a clean API (I wouldn’t really care if it was RESTful or nasty WS-*, frankly). And they’ve been talking about doing exactly that, also for years — the "Case Management API" is the Flying Dutchman of FN promises.

Frankly, they’re also wasting a *huge* opportunity to leverage IBM’s strengths to attack one of FN’s core problems: scalability. The ContentEngine design metaphor starts to buckle when you get up to the millions per day (whether that’s documents coming in, being manipulated, or even just snazzy events being fired). And FN has customers that want to do things at that scale, but can’t. The obvious solution would be to get FN bits running on the mainframe (ideally, natively, but I’d settle for a Linux image). But again, the operative adjectives for their efforts in that direction would be "tepid", "half heated", and "slow".

They argue that enterprises should replace their home-grown bespoke systems for TCO reasons (systems like ones that I’ve written ), but that argument is specious as long as they can’t achieve comparable performance / scalability. The snazzy events that you mention here *are* nice, for example, but *we* told them we wanted that, after we had already gone ahead and implemented it ourselves with bespoke code and MQ Series (where it’s fast as hell).

Now, of course, years later, they are all in danger of being overtaken by events. By the time they get the mainframe proposition sorted, I won’t care — instead, I’ll want to know how they’re going to exploit Amazon EC2, and how they intend to support sharding and Map / Reduce suitable architectures in the CE. LOL.

Financially, they may even be “right” to do so. At least in the short-term. But in the medium- to long-term I think the ill effects of market confusion and increased costs of deployment weigh this approach down.