Peter Lees wrote: > this seems like a fundamental limitation - i'm surprised it's gone > this long without being addressed (i'd actually assumed it *was* > being addressed, otherwise i'd have jumped up & down abotu this > myself, earlier) > > i'd suggest that although the tricky elements described in bug > 6533008 are relevant (behaviour under load, etc), that it would be > worthwhile prototyping this feature rather than squirreling away on a > grand design behind the scenes. getting the feature 'out there' from > a functional point-of-view will considerably enhance SMF's > attractiveness & help spur interest in migrating 3rd party & custom > stuff to SMF sooner rather than later. i say this as someone who's > just been present with entries in /etc/inittab for application > monitoring & restart :P
:) There's no grand design lurking behind the scenes, holding up progress on code. There's just a lack of people to get their hands dirty and start working on the prototype and defining the interfaces. > i'll join a working group, liane, if you think that would help. As with most things in SMF, code is the issue, not ideas. There are a plethora of opinions about what features are important, and how they should be implemented. ;) If you're interested in helping me create and start writing the code in startd as part of an OpenSolaris project, I think we could probably get community approval for an exploratory project. Game? I'm out of time to work much on the project before December (I owe Mark some responses on *his* code contributions, for example!), but am happy to help facilitate and contribute code whenever I can. If the problem is, instead, Sun funding the project sooner, we're probably not in the right forum. (A funded project of this scope would certainly be created in the community -- we sadly just don't have individual engineers assigned to it right now.) liane
