Trying to do two things at once and keep track of the changes to a working system is next to impossible, but this is par for the course for Linden Labs and Oskar cannot be blamed for this. The key fact thoush is not just doing two things at once, but the fact that this is a system at work and you can never tell how it will react to even the best tested code, and if the past is a measure LL code is seldom well tested. My main concern though is introducing new code to the RC servers before we can fully tell the effect of last weeks effort.
Granted last week has let to great improvements in the RC servers, but the memory leaks and the crashes that go with them have not stopped and will not be fully able to be measured til this code is installed on the main servers and we can judge it against the background of the whole system. Right now it is impossible to tell how much memory is lost in a RC server as opposed to a main server and we badly need a week to judge the impact of last weeks changes on the system as a whole and to see the effect today's server work and the work that will inevitably follow will have on the entire system.
Simply, declaring this fixed and rushing on to another project invites disaster and duplicates the behavior that caused this massive problem in the first place, so for once let's err on the side of caution, take a week to access the results of these changes on the system as a whole, clean up the remaining problems then consider what to do next. This I realize is not the Linden way but at this point given the mess we are just emerging from maybe it is time to find a new way based on caution, careful consideration and testing rather than a blind rush toward a lemming like disaster. Again I realize Oskar has little to do with making this decision and am addressing this to his superiors in hopes that they will step back fromt he brink and maybe spare us another disaster that we cannot afford to endure.
