Windows Workflow Foundation
Continuing the discussion from Build process design:
WF follows Fight Club Rules. One does not speak of WF.
But seriously the only reason I know it exists is a project I had to interface with used it.
Link to the rundown on the hellbeast bastard workflow system under discussion is here.
Conceptually, I have basically just fallen in love with this thing. We've been wanting to get away from the XML script bullshit and into a visual designer. Check. We've been wanting to get real flow control instead of the hack-ass bullshit we have. Check.
It would literally be a quick refactor (weeks, not our currently estimated year) to decouple our services from the "service" part and couple them to a custom Activity.
The only real thing I'm still concerned with is scalability. Right now I can add capacity by having my manager shovel an embarassingly large sum of budget to the outsourced datacenter to start up a shitty tiny 2CPU VPM, install a deployment agent, click Deploy, and half an hour later have a linear increase in capacity. Presumably some level of load balancing could be done at the scheduler/initation end, but Wikipedia doesn't cover it.
TDWTF: Ask about fixing your shitty build process, get told how to fix your shitty architecture and look like a Big Damn Hero.
It powers SharePoint 2013 workflows. If you want Workflows (of the 2013 kind) you need at least one Workflow Manager-type of server. http://msdn.microsoft.com/en-us/library/jj193471.aspx
Not that that is really helpful for your real question.
It looks like using Workflow Manager 1.0 limits you to 3 nodes in a (type)farm, but the software doesn't limit the number of nodes added to a farm. I guess one of those wierd "We tested for X and that's all we will support scenarios" but really how often do you pickup the phone and call Microsoft?
AppFabric(based on what I've read) should scale just like IIS so as many machines as you add to the farm should scale out.
Atleast until the persistance(SQL Server) layer is the bottleneck.
So @Weng did you take the WF plunge? Curious because I'm considering using WF for my developers that don't do well with code/documentation/direct-instructions/anything. I figure the pretty pictures they would get would be an improvement over what I got now.
Also, fuck you Ducksauce, I'll necro the shit out of any thread I so choose.
Nope. Still Roll-Your-Own at WtfCorp.
In the final analysis, there wasn't enough benefit to outweigh the risks involved.
Too many features we have that we'd need to hack in, too many more beneficial things we can spend time on. We are building a custom UI, though. Fuck XML as a hand written scripting language. Fuck it right in the asshole.
Tell that to Roku.
If they've got a schema, there are tools that will help a lot. Even without a schema, there are editors that can at least look after closing the entities for you, usually triggering on a key sequence like </.
Doing it all by hand is usually the mark of a masochist.
Based on my last experiences with WWF, the answer should always be No.