I'm looking for technical datapoints on benefits of using wft or wfm one over the other ? Any pointers?
points to consider wfm1. Easy to design the workflow using tool(drag and drop)2. URL task is new in wfm3. Based on branch we can customize values for workflow variables
Just adding one more item to Adam's list: WFT is legacy (still fully supported) technology. Newer features MAY be limited to WFM only in the future. Newer implementations should strongly consider WFM over WFT.
This should be taken with a grain of salt, as I am skeptical whether new features will ever be added at all...
1. The ease of designing the workflow using the WF Modeller decreases with the complexity of the workflow - designing a WFT using the WFConstructor.pm module, IMO, maintains a fairly consistent rate of ease/difficulty
3. However, WFM doesn't [currently] allow you to skip the instantiation screen altogether, while WFT (perl/javascript) does.
I think there's potential for improvements in the WF Modeler UI, and while WFM is going to be Autonomy Interwoven's recommendation for future development, I have to wonder how much improvement will actually ever happen to the WFM GUI (I believe it's OEM'd 3rd-party software). Things I'd like to see: * lock/magnet handles on task boxes (a la Visio / OmniGraffle) so you can make your transition links go from a specific point on one task box to a specific point on another task box and thus help control the transition link line routing. * transition link line crossing options (e.g.: hop-over, bridge-under, etc.) that allow for clearer depiction of which line is "above" another * real task alignment options that understand real grid options * Non-GUI: The ability to skip the instantiation screen altogether (as in WFTs)
Despite what Sales tends to say - the WFM GUI is not meant to be used by non-technical users to develop workflows (it may be used by non-technical users to try to understand a workflow, but not to develop them).
1. The ease of designing the workflow using the WF Modeller decreases with the complexity of the workflow - designing a WFT using the WFConstructor.pm module, IMO, maintains a fairly consistent rate of ease/difficultyI have only found this to be true if you have more than 3-4 transitions to/from a single task. The only scenario I can recall is a task I had that was transitioned to from any external task that errored or timed out and could also transition to any task in the workflow. I have since refrained from doing this (although it was nice) and all of the workflows I've developed since look very good in the modeler. I do admit, the system I now use for laying out tasks so that they are easily readable took a while to work out, but now seems to fit most of my needs.
This [trying to skip instantiation screen]is annoying, yes, but can be done with some simple (unsupported) javascript.
After playing around with WFM a bit, I haven't had too much of an issue with the positioning of links, you can get them to look ok once you get the hang of it (but the fact that you need to work at it is annoying).
When I first started using WFM, I put together a list of things I'd like to see (although likely never will):* Be able to specify the order of successorsets. They currently appear to be random. I had once thought they were controlled by the order the links were added and/or the order they appear on the task, but have since revised my opinion. I have seen workflows in production that change the order of the successorsets all the time, without any code ever changing. I used to rely on my error transitions being successorset 0, but I now do all of my transitions based on the name of the successorset, rather than the index.
...* Have an option to output the IPM to Visio.
Oh well - we can but wish. I'll be doing some WFM development later on in my current project (672SP2P1) and I'm sure I'll run into some other frustrating things ;-)
I too used to use 0 as my error handling transition, but now I just name all such transitions "error" and it works just fine. If you use Perl code - the WFTask module (as of 6.7.1 I believe) provides the GetPossibleTransitions() method which returns the array of transition descriptions - from which you can find the one you want and use it's index in that array to do the CallBack()
I assume you're providing this for the benefit of future readers? (I mentioned I use the successorset label as well in my post ).
But, it is still annoying for user and group tasks, when the transition buttons in the task view don't have a fixed order.~Jeff