Starting by running the tutorial. Working around issues like:

• can't use CI_HSC_DIR environment variable with lsstsw installs.
• PS1 reference catalog changes the set up directions.
• Can't assume that MPI will work for everyone out-of-the-box with singleFrameDriver.py

We'll have to avoid -j 2 or greater with processCcd.py in the quick start until the DM-10834 issue is resolved.

John Swinbank, could you help me find a suitable technical reviewer?

Also, Jim Bosch, although you don't have to give a full review, I'd be interested in your take on whether this tutorial series will be useful for the workshops you're giving. I'd like to add additional parts to the tutorial series on image display and source analysis.

Summary

• This PR adds a new "Getting Started" section on the homepage. This section is meant to give a good onboarding experience for new and prospective users.

• The tutorial itself is adapted from https://dmtn-023.lsst.io and currently focuses on using command line tasks to ingest data in the Butler, using processCcd.py, making coadditions, and measuring sources. The series can be expanded later to include image display and table analysis tutorials later.

• An installation tutorial guides the user through the installation choices and makes sure the user has Git LFS set up to clone same ci_hsc data.

• Added a small Git LFS configuration document that's oriented towards end users (as opposed to the more complex document at https://developer.lsst.io/tools/git_lfs.html).

• Refactored the system macOS and Linux prerequisite from the newinstall.sh page into separate to make it easier to follow the newinstall documentation. More work is needed on that documentation to separate the "how-to" from the reference aspects of the newinstall installation.

Note: a blaring issue with the tutorial is that the installation points to the tag of the last major release, while the tutorial itself likely needs a more recent release. This can only be worked out with a proper release and CI process for the documentation. This needs to be worked out in follow-up tickets.

New pages

• https://pipelines.lsst.io/v/DM-10637/getting-started/installation.html
• https://pipelines.lsst.io/v/DM-10637/getting-started/data-setup.html
• https://pipelines.lsst.io/v/DM-10637/getting-started/processccd.html
• https://pipelines.lsst.io/v/DM-10637/getting-started/coaddition.html
• https://pipelines.lsst.io/v/DM-10637/getting-started/photometry.html
• https://pipelines.lsst.io/v/DM-10637/install/git-lfs.html
• https://pipelines.lsst.io/v/DM-10637/install/prereqs/index.html (and linked pages)

Updated pages

• https://pipelines.lsst.io/v/DM-10637
• https://pipelines.lsst.io/v/DM-10637/install/newinstall.html
• https://pipelines.lsst.io/v/DM-10637/install/lsstsw.html

The workshop I was giving has already happened, and the topic for that tutorial was running slightly lower-level Tasks (essentially the leaf subtasks of ProcessCcdTask) from Python, rather than running higher-level processing from the command-line. But I pointed the attendees at DMTN-23 as something they could work through on their own to cover the latter topic, and around this time last year I did use the first draft of DMTN-23 for a workshop. I don't see any reason why your version wouldn't work at least as well for both of those purposes, once it's cleaned up and merged. I think a workshop based on it would probably work best as a session in which attendees worked though it at their own pace and got help from roving experts, which is more or less what I did last year. This year's tutorial (which had materials in Jupyter notebook form) was much more interactive in that it had some exercises for attendees to work on between more lecture-like steps. I think that worked much better for a live tutorial, but I think it'd be a lot harder to have it also serve as standalone documentation.

My suggestion would be that over the course of this week you discuss this tutorial with Lauren MacArthur and have her read over it. If she's happy, I think we can regard it as reviewed.

Hey, Jonathan Sick, Lauren MacArthur —

Since there's a lot of work involved here, Frossie Economou & I suggest that we regard this ticket as done, and file a new ticket which is essentially "work through the tutorial and check that it all works", which, I gather, is what Lauren MacArthur has effectively been doing. Does that work for you both?

Jonathan Sick, I'm not sure what your plans are for putting this live on pipelines.lsst.io — my suggestion would be that we wait until after the follow-up ticket described above.

That works for me

OK. I guess I can put this on a user branch for integration? (for what it's worth, combining business accounting with the Git branching has its downsides)

It would certainly be better to merge this after v14 is out (the tutorial won't be compatible with the 13 or 13.1).

DM-11077 and DM-11034 are related tickets that add additional tutorial steps. Should I treat those tickets the same way? Close for now and add them to the new review ticket?

For fear of them getting lost, I submitted some preliminary comments on the PR.

Thanks Lauren MacArthur!

Sorry for being slow on getting back to this.

OK. I guess I can put this on a user branch for integration?

(for what it's worth, combining business accounting with the Git branching has its downsides)

I think that's partially true, but you yourself raise another issue here: we don't want to merge to master until after the next release.

(Actually I'm not sure how that'll work in general — will we tag pipelines_lsst_io at the same time as everything else which makes up the release? And show the version corresponding to the most recent release, rather than master, by default?)

DM-11077 and DM-11034 are related tickets that add additional tutorial steps. Should I treat those tickets the same way? Close for now and add them to the new review ticket?

I think the same approach, but maybe making a separate "review" ticket for each one, rather than combining them all into one.

Thanks for your help with this!

Closing this ticket out. I've created a new branch and PR with DM-11391 (I didn't go with an integration branch after all; instead DM-11391, DM-11392, and DM-11393 just branch off each other).