Closed bartelink closed 4 months ago
I'm definitely not trying to push this one as aggressively as 184, but I am pretty confident that there's nothing to be gained by having a net7.0 or net8.0 build of the package
Obviously if it's possible to achieve more or improve performance by targeting an API only present in a newer TFM, then there's a choice to be made between:
net6.0
My objection is to making life more complex than it needs to be until such time as there's something in a newer TFM that's relevant to the task at hand.
potential follow-up to #184 See https://github.com/serilog/serilog/issues/1970 for lots of reasoning, but the TL;DR is:
In the general case, a library should depend on the minimum TFM that it can be implemented in terms of. For many libs, thats
netstandard2.0
, and supporting that is great when it's possible.In the
netstandard2.1
-net5.0
timeframe, there are various confusions, and ultimately there's not benefit to specifically targeting them or thinking about themnet6.0
brings significant available API surface and removes the need for complex conditional dependenciesAdding TFM-specific builds for stuff later than
net6.0
on the other hand brings little concrete benefit. i.e.net9.0
is "supported"