* a number of projects in the train, including Platform, depend on ECF, so we may end up with different projects pulling in different versions of ECF, which is not ideal
* The fact that participation in the train entails so much overhead that an active project with a long history of successful participation is considering pulling out seems like a larger and quite serious problem for Eclipse. I believe at least part of the overhead is due to the instability of the build infrastructure.
I'm afraid I don't have any concrete solutions to propose, but I wonder if, given the current resources, we should consider simplifying the contribution process (somehow), or reallocating resources from elsewhere to improve the stability of the infrastructure.
Regarding the infrastructure, my experience from the Mylyn project is that projects must spend a great deal of time dealing with broken Hudson slaves, breakages in the signing service, and similar issues, to the point where I can see why it threatens projects' ability to participate. Just to be clear, I am not blaming anyone for this - I know that it takes a lot of work to keep all these things running smoothly.