Hi <!subteam^S016E36N7TQ>, I’ve spoken to <@UE4L64...
# pipeline-maintainers
s
Hi <!subteam^S016E36N7TQ>, I’ve spoken to @Phil Ewels about being assigned an admin role in nf-core/oncoanalyser so that I can manage branch protection rules for specific development branches - there was recently an instance where a user inadvertently made unannounced changes to a branch under active development that caused us some headaches
s
hi Stephen, could you share the branch where those unannounced commits were made ? (asking purely for learning purposes and to see if we can put up some guardrails if needed)
s
Sure, the branch is pipeline_v6.0 - I have since revert the changes by interactively rebasing the git history locally and dropping the unwanted commits then force pushing
And maybe related to your line to thinking: I would find it most useful to have the ability to readily add (or potentially remove) users that can write to a protected branch that is under active development rather than having a somewhat static list of permitted/trusted developers
e
@Stephen Watts Check out https://github.com/nf-core/ops/pull/59 It would allow you to make a PR to do that and then a core team member could merge it
Aside: What's going on with all of the pipelline_v6.0 branches? Are these on purpose?
s
Thanks @Edmund that’s a pretty interesting approach, really like it! Looks to be some way off though, I’ll stay tuned for updates 😃
Aside: What’s going on with all of the pipelline_v6.0 branches? Are these on purpose?
The development involves early-access for several groups who are testing patient / research data to give feedback on performance and accuracy of analysis, and so we need to have stable and human readable pointers to specific points in history for these groups to access
All
pipeline_v6.0.*
branches will be deleted once we are satisfied with how the pipeline is performing and the changes then merged into
dev
for a full nf-core release
It’s a little noisy but it really helps with provenance and debugging across several groups rather than attempting to deal with a single, often changing branch
e
Hmm that’s an interesting approach. So is the new version going to be 6.0.0? Would you be open to using normal fit tags for these rc releases?
s
The next version of oncoanalyser will be 2.0.0 - the v6.0 is coming from the workflow it largely implements, WiGiTS
I would love to be using GH ref tags for these release candidates but I have been told that the tags are reserved for full nf-core releases
And definitely open to hearing more of your thoughts on improving or alternative approaches that work within nf-core
Outside of nf-core and what I understand of the expected ops / rules, I would absolutely be using git tags and GH pre-releases
Maybe we can continue discussing this somewhere else since we might be straying into off-topic territory for the thread