SPEC 0: Soften explicit drop schedule.

[Carreau]

This does only change the presenting of the drop information:

• reduplicate when the same package was released twice in the same quarter.
• Remove the day of month. I don't think it helps conveying the flexibility that the spec

--

This also regenerate the files

[jarrodmillman]

The format of spec-0000/schedule.md needs to be fixed. It doesn't pass the linting and I suspect the rendered text isn't what you intended:

You may want to use the yamltotable shortcode:
https://theme.scientific-python.org/shortcodes/#yamltotable

[Carreau]

Yeah, I'm trying 4 spaces to have pre blocks, it's not the best.

I'm going to see if it supports tables.

[munkm]

Remove the day of month. I don't think it helps conveying the flexibility that the spec

I think this is a good idea and I'm in support of this change.

Something I've been thinking about with respect to the flexibility of this SPEC is the implications of the word "drop". I wonder if we add language saying that "drop" doesn't mean explicitly removing support, but rather not continuing support for that versioning. What do you think?

ETA: Whoops. I see that conversation is already happening in #272

[Carreau]

Better as a table:

[bsipocz]

I really like the cleaner look of your table solution than what we currently have.

[bsipocz]

I feel that the previous quarter could be kept around for a bit of an overlap.

[bsipocz]

The rendered version could maybe benefit from a tiny bit of different formatting, either leaving more space between this line and the table itself, or making this line with a different typeface (italic should be enough, no need to make it a heading).

[Carreau]

Yeah, spacing need css change, as prettify reformat the markdown anyway.

[bsipocz]

what about #273 (comment) then?

[jarrodmillman]

You can put the markdown inside of

<!-- prettier-ignore-start -->

and

<!-- prettier-ignore-end -->

to prevent prettier from messing with your formatting.

Feel free to let @alphapapa know if you would like any css updates to the theme. You are also welcome to just make a PR if you prefer.

[Carreau]

I've tried reading 1 quarter in the past.

[bsipocz]

I quite like what I see under the current preview link, not sure why prettier doesn't like it though.

[stefanv]

Easiest is to run the generated result through prettier and look at the resulting diff.

[jarrodmillman]

Or ask the pre-commit bot to autofix it.

[stefanv]

If it's for a generated file, you'd have to run that each time you produce the file.

[jarrodmillman]

Here is what prettier does. For example, it changes this

|    |    |    |
|----|----|----|
|ipython        |7.26.0 to 7.28.0   |released Aug 2021 and Sep 2021|
|networkx       |2.6                |released Jul 2021|
|pandas         |1.3.0              |released Jul 2021|
|scikit-learn   |1.0                |released Sep 2021|
|xarray         |0.19.0             |released Jul 2021|

to

|              |                  |                                |
| ------------ | ---------------- | ------------------------------ |
| ipython      | 7.26.0 to 7.28.0 | released Aug 2021 and Sep 2021 |
| networkx     | 2.6              | released Jul 2021              |
| pandas       | 1.3.0            | released Jul 2021              |
| scikit-learn | 1.0              | released Sep 2021              |
| xarray       | 0.19.0           | released Jul 2021              |

I would update the script to just generate the table so that prettier is happy. It might be easiest to use something like:
https://pypi.org/project/tabulate/

[jarrodmillman]

There is a blank line at the end of schedule.md that I still need to fix. After that, this should be ready to go.

[jarrodmillman]

@Carreau @bsipocz I removed the final blank line in 201e923. The CI is passing now and the tables are formatted in the new style. I am sure the code can be improved. Feel free to fix my commits or just delete them and fix the PR in a different way.

[bsipocz]

The rendering looks good to me so I keep my approval for merging as I don't have strong opinions on how the code archives that
(I have the feeling that the astropy fixed width ascii table writer could recreate something like these in much fewer lines, but don't think it would be worth the effort of rewriting everything)