docs(blog): portable dataflows with Ibis and Hamilton

[zilto]

Description of changes

Ported a blog post from the Hamilton Substack to the Quarto format for cross-posting on the Ibis blog. This was discussed with @lostmygithubaccount on Zulip.

First time using Quarto. I managed to successfully build .qmd files and preview the post. Everything seemed fine. However, I see other blogs have files under _freeze/posts/, but I couldn't find how to generate them.

[github-actions]

ACTION NEEDED

Ibis follows the Conventional Commits specification for release automation.

The PR title and description are used as the merge commit message.

Please update your PR title and description to match the specification.

[lostmygithubaccount]

thanks! I think this will fail the docs PR check, as hamilton will not be installed in the environment. you'd need to render the document yourself with quarto (via quarto preview or quarto render) and check in the changed _freeze files. this will have quarto use the saved execution output for the blog instead of trying to run it

[cpcloud]

@lostmygithubaccount It's passing because the code blocks aren't executed.

[lostmygithubaccount]

ah missed that -- will review soon! @zilto you may want to have the code blocks actually execute (and then my comment above would apply)

other offhand comment before a proper review would be wrapping the lines to ~90 characters -- see https://ibis-project.org/contribute/03_style#text-formatting

[zilto]

@lostmygithubaccount It's passing because the code blocks aren't executed.

aaah, that makes a lot sense. I was scratching my head at why no _freeze/posts/hamilton-ibis directory was generated.

Will make the changes

[zilto]

I updated the .qmd file to use executable code blocks and generate the _freeze/ file.

Now, my code examples depend on a parquet dataset that is 23KB. I didn't commit it since .parquet files are git ignored by default. Should I include it to make the code runnable?

[lostmygithubaccount]

@zilto apologies for the slow response here!

Now, my code examples depend on a parquet dataset that is 23KB. I didn't commit it since .parquet files are git ignored by default. Should I include it to make the code runnable?

I think this is small enough that just adding it is fine. alternatives would be:

1. host this somewhere publicly that can be read from
2. switch to using a built-in ibis.examples dataset

neither really seem worth it in this case, and 2) would mean changing a lot of the code

[gforsyth]

@zilto -- if you add the parquet file here (or you can send it over Zulip to me -- at least I think that's possible) I can stick it in one of our buckets and give you a public http link to pull from.
I agree checking it in is not a huge deal, but it's also easy to stick in the bucket and have a reasonably permanent link (in the medium-term, anyway)

[lostmygithubaccount]

it is trying to run the Python code (and not finding the hamilton import), suggesting the execution freeze doesn't exactly match after you made some changes. can you try re-executing and ensuring it's up to date?

[zilto]

Ok, odd bug... I just updated the files

[ibis-docs-bot]

Docs preview: https://pr-8798-a34e1ad2b7a39f14e9f7b0025ac68c3496402e00--ibis-quarto.netlify.app