export Logger.Sink #66
Conversation
pohly
force-pushed
the
export-sink
branch
2 times, most recently
from
5d2a998
to
105badb
Compare
|
This seems good. Before I merge - can we think of any way this could cause problems or compat breaks? I don't think so, but these things are subtle. |
|
That's why I filed issue #67 and referenced the apidiff testing from klog 😁 I had a quick look right now and tried to run apidiff manually, but must have done something wrong because it didn't even warn about some intentional API break that I introduced for testing the tool. FWIW, I also think the change is safe. |
|
apidiff doesn't seem to work with Go 1.16. I got output only after building it with Go 1.13.3. This dampens my enthusiasm for the tool a bit - I had expected it to keep up with changes in Go. But once it runs, it seems to work. It correctly detected the intentional API breaks and for this PR reports: |
|
hm it seems to work for me. let me see. |
This is necessary to support the "breaking class" use cases. One example for that, access to the underlying implementation, gets implemented for funcr and tested with a runable example.
This is necessary to support the "breaking glass" use cases. Two
examples for that, access to the underlying implementation and a
custom With* function, get implemented for funcr and tested with
runable examples.
Fixes: #63