On Mon, Aug 20, 2007 at 03:46:07PM -0400, Richard Lowe wrote: > Bill Sommerfeld <sommerfeld at sun.com> writes: > > > if by "access list" you mean the set of people allowed to change the > > state of the gate, shouldn't the access list really be the union of: > > - the gatekeepers > > - the set of people with approved RTI's? > > When the gate moves to opensolaris.org, the RTI system and the gate > cease being able to communicate in any way shape or form > (hg.opensolaris.org is not on SWAN). > > However, when that can be rectified, that's an idea I really like, and > should certainly be implemented. > > All assuming I should read it literally, as in "You may only putback iff > you have one or more already-approved RTIs".
I'd really much rather see a change accepted or denied based on whether that change was approved or not, rather than whether the RE had any approved RTIs. It's simply not restrictive enough (i.e., an RE with two RTIs, one approved, the other not, could push both changes if you take Bill's approach). As for having the RTI system outside, we should at least be able to start work on having *an* RTI system outside. Can we at least release the code so that those with strong itches can start drawing blood? But more useful is having at least a subset of the RTI database outside (I'm not sure how much of it is proprietary, or could be considered so). At least the RTI checking bits of cadmium and the gate machinery don't need all the information in the database, and we could do worse than limp along by pushing a subset of the data out every few minutes, or on each update. It wouldn't be enough to allow for non-Sun advocates, or even to submit RTIs from outside the firewall, but it *would* allow for putbacks. Danek
