Philip Brown wrote:
Dave Miner wrote:
Darren Reed wrote:
My fear is that we are creating too many FMRIs and that
in doing so, we're creating problems for ourselves in
the future. Current FMRI count is closing in on 300 in
Nevada. Nearly 3 times the number from S10FCS.
Fear not. To the contrary, the number of services on the system
indicates success on a number of fronts:
- Deeper integration of SMF into the system
> ...
"use" is not always synonymous with true "integration". I think that
Darren might be concerned with the possibility that FMRIs are being
used, simply because there is no cleaner way to do certain things,
rather than
"yes, these 300 things are really best done as 300 individual,
separate FMRIs"
The sheer number seems to hint that perhaps a review of them is in order.
Right.
And whilst PSARC reviews them one by one, my fear is
"death by 1000 cuts" - each one by itself seems innocuous,
but if you step back and see the bigger picture...
does the view change? My fear is that it does but
maybe I'm jumping at shadows.
Definately Dave Miner has a point in that a lot of the
growth in SMF FMRIs is because of things migrating to SMF
since S10FCS but I can't help but wonder if SMF is our
hammer and we're trying to turn everything into nails?
For example, rather than use SMF FMRIs as the actuators
that do file assembly, why isn't that an integral part of pkg(5)?
Why isn't pkg(5) providing the file assembly through its own architecture?
Or maybe a better question is, why can't pkg(5) provide that?
If there is a limitation, is it architectural or from the implementation?
Or is it just a design choice?
Darren
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
