I have a general question around how do you onboard your colleagues at work when you introduce a new tool e.g. pants in my case. I wrote a wiki and had a meeting/tutorial, but I still feel like that people will be bit lost when they start using it. I work in a data science team and I feel like not everyone is so inclined to handle the engineering complexity that comes with introducing such tools.

You're describing my situation as well.

This is a great question. We want to create more documentation around this on pantsbuild.org. But even that is probably more than regular users want. The documentation is really more for the people driving adoption, such as yourselves.

I think a meeting/tutorial/workshop covering 80% of what they "need" to know is likely the best option. After that, linking to the docs or Slack and then providing some initial support as well (before the users land here) is probably just something that'll need to exist. (Not because of Pants, but just because of the scope and nature of the change)

How about a simple list of recipes: " To run a test:

./pants test path/to/test.py

To run all tests:

./pants test ::

To run linters

./pants lint ::

"

yeah @happy-kitchen-89482 the wiki I worte tries to introduce them with such simpler concepts and that's how I sold the usage of pants to begin with in my team :)

Things that people can just copy-paste into a terminal without thinking too hard

To tack on to @happy-kitchen-89482’s I think you could provide "when you used to do X, you now do Y"

I remember trying to learn git, even Pants is simpler than that...

E.g.

bazel run //foo/bar

is now

./pants run foo/bar

, etc...

but in 2010 it was so confusing, coming from Perforce

Of course now everyone "understands" git

Fixed that for you

I think those who have had experience with any build tool like say makefile, can grab the concepts around pants bit easily. But I doubt lot of data scientist ever used any build tool before.

@refined-addition-53644 we're coming from Bazel, so it's a really easy mapping/sell. I guess you're starting from nothing?

yeah we went from poetry (which was already new for some data scientists) to pants

Ahh, so still relevant then: • If you were running

poetry run pytest ...

now use

./pants test ...

• ...

I suppose I actively need to make some team members to try to add a new dependency and then run all pants commands like

tests

,

fmt

etc. Then when things fails they can either figure it out themselves or I can give a hand.

That's why I think a Pants workshop is a good idea as well. E.g. "everyone follow along with me"

BTW we (Toolchain) can be available to give talks, workshops and trainings