MiddlewareErrors really shouldn't be in syslog for non-debug builds
We're already short on space for /var . Having [[MiddlewareErrors]] be reported in syslog for valid operations (that should be trickled up the stack) just seems like a really bad idea, and this will blow up release builds that have been up and running for months where transient, correctable errors are being reported on a periodic basis.
#2 Updated by Anonymous over 9 years ago
Replying to [comment:1 william]:
Do you really think that is a bad idea?
Otherwise there is no other idea yet to view errors that happened in the past
TBH the argument about short on space for /var is very weak, those errors are not frequent and would hardly eat up more than 10k
Ok then. We can leave it on for the 8.2 release, but really... we need to revisit this item in the future once we have a sane logging subsystem. It discourages me from using this because I like using Exceptions in python code to trap unwanted scenarios and trickle it up the stack, instead of worrying about testing return values, etc. Using well defined Exceptions will help us create better APIs that will withstand the test of time, instead of continually rototilling things, and thus will help us design a more robust platform.