Ah. That is illustrative, Bryan. One of the most attractive things about the ACS (at least to me 2 years ago) was its responsible handling of persistence, namely handing it off to Oracle; in which the ACS stood in stark contrast to its competition in Vignette, Broadvision, Dynamo, & co with their homebrew, unstable, difficult to administer caching & storage systems. ACS-tcl was a more-or-less stateless service layer. Down the latter path is not *one* "server crash" issue, but a lifetime of them. I fear that aD may be, after taking steps forward in flexibility and API orthagonality, taking a solid step back in reliability.
I certainly could be incorrectly assuming too much of aD's future architecture plans, and thus could be spectacularly wrong. Time will tell.
