The motivation for this Issue is an attempted candidate build I tried to perform over the weekend, https://github.com/DUNE-DAQ/daq-release/actions/runs/8868482226. The immediate cause of that build failing is that hsilibs couldn't compile, but the reason for that is that hsilibs, despite beings listed in the relevant release.yaml as being v4.1.0didn't yet have that tag. Indeed, on line 433 of the output you see
Info: hsilibs | v4.1.0 | b6e0904
...where what's noteworthy is that the commit b6e0904 represented the head of the production/v4 branch at the time (which aligned with v3.1.0, modulo some Workflow *.yaml changes).
In such a situation - i.e., there's not actually a git tag in the repo corresponding to the requested tag - make-release-repo.py should immediately exit with a failure. We were lucky that hsilibs didn't happen to compile, but this falls under the category of "Better to fail loudly than silently and unnoticed".
The motivation for this Issue is an attempted candidate build I tried to perform over the weekend, https://github.com/DUNE-DAQ/daq-release/actions/runs/8868482226. The immediate cause of that build failing is that
hsilibs
couldn't compile, but the reason for that is thathsilibs
, despite beings listed in the relevantrelease.yaml
as beingv4.1.0
didn't yet have that tag. Indeed, on line 433 of the output you see...where what's noteworthy is that the commit
b6e0904
represented the head of theproduction/v4
branch at the time (which aligned withv3.1.0
, modulo some Workflow*.yaml
changes).In such a situation - i.e., there's not actually a git tag in the repo corresponding to the requested tag -
make-release-repo.py
should immediately exit with a failure. We were lucky thathsilibs
didn't happen to compile, but this falls under the category of "Better to fail loudly than silently and unnoticed".