Having had time to reflect a bit, I'm not sure the 'dh-go' approach
really gives us what we want.  Wouldn't the following process work?

1. Upload dh-golang in experimental with fixes.

2. Rebuild all Go packages in unstable, pinning the dh-golang from
experimental, through some archive-wide rebuild person.  (Uploading
packages to experimental is a working but non-scalable substitute.)

3. On FTBFS's that doesn't happen in pure unstable, fix things either by

3a) Package upload into unstable that makes the package works BOTH under
the dh-golang in unstable and dh-golang in experimental.

3b) Some patch to dh-golang in experimental.

4. Goto 1 until there are no more FTBFS.

Unless I'm missing something, then having a 'dh-go' in unstable doesn't
really help a lot.  Unless it is somehow part of the 3a) fix?  Maybe
that's the answer.  But I don't know how any 3a) fixes would look like.
Have we made any upload like that so far?  Beyond the go.mod-adding
uploads.

/Simon

Attachment: signature.asc
Description: PGP signature

Reply via email to