On Thu, Sep 29, 2011 at 4:45 PM, Nico Williams <[email protected]>wrote:
> On Thu, Sep 29, 2011 at 3:28 PM, Jeff Victor <[email protected]> > wrote: > > The general rule is "convince product management that there is a business > > reason to invest the engineer(s) and it will get done." > > IMO, for backports, the bar should be much higher. Yes. I intended to say "sufficient business reason" to cover all that. My bad. --JeffV > The vendor should > compute the cost of the backport *including* the cost of opportunity, > and including the further cost of opportunity involved in encouraging > more backports by the mere fact of having done one backport (if the > customer believes they can put off upgrading forever then the pressure > to backport more and more features will rise). If the value of doing > the backport *significantly* exceeds that cost, then, sure, do the > backport. > > The cost of backporting complex features, particularly ones that have > wide ramifications, and particularly when the backport is to Solaris > 10, with its awful patching mechanisms, is best understood as > astronomical. A backport of Zoned NFS server should be considered as > in the high 7 $ figure range, if not higher still -- after all, how do > you estimate the forgone value of talented engineers working on > innovative new features?? > > Just say no to backports. Pressure the ISVs instead to re-certify > their apps. Legacy costs the customer a lot also -- there's enormous, > typically unaccounted-for costs in legacy. > > Nico > -- >
_______________________________________________ zones-discuss mailing list [email protected]
