On 2021-6-2 23:39 , Christopher Nielsen wrote:
I absolutely support the package/namespace “named logger" approach. That
concept is also used in the Java/.NET worlds, as well as others, and
it's incredibly useful (and powerful). And since we’re already taking
advantage of TCL namespaces, that provides a straightforward segregation
scheme.
And ultimately that would be the way to go, as we evolve our logging
support.
Short-term, though, I was hoping we could start with the simpler
approach: Expand our debug granularity, to filter out less-useful info
by default, but without removing the detail entirely. Indeed, there are
some areas where I’d like to see /more/ detail - albeit, logged at level
debug1 or lower - to help with troubleshooting.
This isn't an urgent change. Let's finish working out the design before
we merge an implementation. To do otherwise is to invite technical debt.
- Josh
