I’ve just finished the first version of DMTN-23, a tutorial on using LSST command-line driver tasks that I’ll be presenting at the LSST@Europe2 meeting this week. It uses the data in ci_hsc, but walks the user through running all of its steps manually (and it uses the higher-level pipe_drivers tasks instead of calling the ones in pipe_tasks directly).
The live tutorial will probably be over before most of you start your work day, but I’d like this document to continue to serve as one of the first documentation entry points in the future, as I think it’s currently our best documentation on how to run what is essentially the HSC survey pipeline through the LSST side of the stack.
To that end, I encourage all Science Pipelines developers to read through it when they get the chance, and ask that they open issues to change anything they think is incorrect, misleading, or flagrantly missing. I’d also very much like help in keeping it up to date - there are several places where the tutorial provides workarounds for “non-ideal” behavior that I expect to be fixed in future versions of the stack (I’ve recently opened some issues for some of these), and I’m hoping we can all be aware enough of the tutorial content (both as implementers and reviewers) to update it when that happens.
I’d of course be quite happy to eventually see this moved to another location that’s more integrated with other documentation (maybe https://pipelines.lsst.io is ready for this?), and if there are other pieces of documentation that are redundant we should merge them or deprecate one. In particular, it’s not clear to me how much the Confluence SWUG page on stack tutorials is still being updated, so I’m hesitant to do anything myself there right now.
I really appreciate that you took the time to write this tutorial. I’d encourage others to write other tutorial technote as well. It will be very easy to migrate this content to https://pipelines.lsst.io, the future documentation site of lsst_apps, which will directly replace both the Doxygen site and the SWUG. There’s no really issue with creating technotes now that will be deprecated and refactored later.
Other parts of DM will have, dax.lsst.io, qserv.lsst.io and firefly.lsst.io documentation sites, for example.
Again, thank you for contributing this.
As part of the software documentation sites, like https://pipelines.lsst.io, we will have a framework for testing tutorials in a CI environment. This will ensure that tutorials and examples in the docs are always runnable against the current version of the software.