Markdown

[benjamin-kirkbride]

Accidentally closed #1314 😅

Features:

• indenting and dedenting of list items
• pressing enter with a list will continue list at current indentation level
  • support task items
• pressing enter on an empty list item will delete list item
• continue task lists
• continue > quotes

Closes #1305

[Akuli]

As for git messes: it's totally ok to make a mess with git! I use "squash and merge" in my projects, which basically means that a messy PR will appear as just one tidy commit when someone looks at the Git history.

You might also like https://akuli.github.io/git-guide/

[Akuli]

I like it overall :)

[Akuli]

Can we keep this, but just modify it so that it passes?

[benjamin-kirkbride]

I got rid of it because it felt redundant with all the markdown tests, but I can look back and see if there is a way to make that happen

[Akuli]

These are maybe a bit too thorough :) We really don't need more than a thousand green dots when running pytest because of one regex. But more importantly, these two lines aren't very readable. It took me a while to figure out what they do.

Maybe instead add a few specific cases, so that it's easier to understand?

ListItemCase(id=f"numbered 1", line=f"1. item 1", expected=True)
ListItemCase(id=f"numbered 1", line=f"1) item 1", expected=True)
ListItemCase(id=f"numbered 4", line=f"4. item 1", expected=True)
ListItemCase(id=f"numbered 4", line=f"4) item 1", expected=True)
ListItemCase(id=f"numbered 42", line=f"42. item 1", expected=True)
ListItemCase(id=f"numbered 42", line=f"42) item 1", expected=True)
ListItemCase(id=f"numbered #", line=f"#. item 1", expected=True)
ListItemCase(id=f"numbered #", line=f"#) item 1", expected=True)

Also applies to other .extend()s.

[benjamin-kirkbride]

We really don't need more than a thousand green dots when running pytest because of one regex

But it's so fun 😩

these two lines aren't very readable

That's fair. It felt readable at the time 😓

Maybe instead add a few specific cases, so that it's easier to understand?

Yeah I can do this.

On a more broad note, how would you feel about adding https://hypothesis.readthedocs.io/en/latest/ to the test suite?

[Akuli]

I'm not a big fan of hypothesis. IMO tests should be as concrete/specific as possible, such as ListItemCase(id="numbered 42", line="42) item 1", expected=True), whereas the code being tested tends to be written very generally. This ensures that at least the test is correct, because it's ideally very simple.

And yes, I know that the point of hypothesis is to catch bugs where a specific string that you didn't think of causes a problem. But they tend to get caught without unit tests in Porcupine. For example, the empty line.splitlines() bug caused errors when I opened a file.

I can see hypothesis being useful in other projects though. For example, maybe you have functions to escape and unescape a string in some way, and you want to know that they work in all corner cases.

[benjamin-kirkbride]

I would see hypothesis as purely additive, it wouldn't replace the simple ListItemCase(id="numbered 42", line="42) item 1", expected=True), sanity type tests, but it would replace:

IS_LIST_ITEM_CASES.extend(
    [
        ListItemCase(id=f"numbered {i}", line=f"{i}{sep} item 1", expected=True)
        for i, sep in itertools.product(itertools.chain(range(100), "#"), (".", ")"))

For example, the empty line.splitlines() bug caused errors when I opened a file.

I think there is likely many more of these sorts of things, especially as more and more plugins get added ;)

I think hypothesis would likely help a lot with the flaky tests as well. I should probably move this discussion to an issue.

[benjamin-kirkbride]

@Akuli does the error in CI mean anything to you?

[Akuli]

The CI error has to do with #1300. Apparently the CI on Python 3.11 doesn't pass very reliably, even though you got it to pass most of the time in #1327.

I'd suggest we do nothing about the CI error if it disappears when tests run again. We will "soon" switch to pyright anyway.

[benjamin-kirkbride]

This PR is blocked by #1349

[benjamin-kirkbride]

I am now going to work to finish this using Actions