Date order on main page?

[defjaf]

Would it be useful to implement a “date ordered” list on the public-facing PEP page? I understand that "date" is ambiguous, but what I have in mind is something more-or-less like the numerical index, but with the various special cases (3xxx, 8xxx, 666, 754, ...) inserted when they were initially created (or some other relevant date TBC). This would make it easier to see the most recent additions and, not incidentally, make the history somewhat more transparent.

[Rosuav]

"Creation date" isn't ambiguous, so that would make a reasonable definition. I don't think any other date would be useful enough to consider.

Since it's a git repository, the creation date can be ascertained by finding the commit that introduced the file. Something like this:

ls pep-*.txt | xargs -n 1 -I@ git log --diff-filter=A --follow --format='%aI @' -- @ | sort

It takes a while and does a lot of chugging, and ultimately, the result is very similar to a numerical listing. I'm not sure how valuable it would be to have yet another index in the same page.

What might be more useful, instead, would be a JavaScript-managed page that allows people to sort the PEPs by whatever they choose, with all the sorting happening on the front end. Not sure whether it's worth the effort, but it would certainly be cleaner than having myriad pre-sorted lists in a long page.

[encukou]

Alternatively, it could be on its own page. Maybe the Numerical Index could move to a separate page as well.

[CAM-Gerlach]

Even better, we can just make the tables, particularly the numerical index, sortable by header (as @Rosuav suggests), and show a few additional useful headers (Creation date, Track if and when that gets added, and perhaps split out the status and type headers into columns. That way, people can have any sort they want in a single table, with a single click. It shouldn't really be that more complicated than generating a whole separate page wit ha different sort order, but much cleaner, more user-friendly and functional.

[brettcannon]

Another option is also collapsible tables which can visibly hide tables that are typically not used. That would allow for potential styling differences between the tables as appropriate compared to trying to make it all fit into a single table format.

I have no clue how important any of that is, to be clear. 😁 When I go to the index it's to find a PEP whose number I can't remember.

[CAM-Gerlach]

Another option is also collapsible tables which can visibly hide tables that are typically not used. That would allow for potential styling differences between the tables as appropriate compared to trying to make it all fit into a single table format.

Yeah, I'm not necessarily suggesting eliminating the other tables, just adding sort functionality to them (and ideally a few more columns to at least the big "sort by index" one). Collapsing tables would help greatly reduce the impact of adding additional ones (for whichever reason) and make the existing page easier to navigate. In particular, I'd think the author index could be collapsed by default, as I'm not sure if anyone actually uses it.

I have no clue how important any of that is, to be clear. grin When I go to the index it's to find a PEP whose number I can't remember.

Yeah; based on feedback on this related thread, I'd wager the large majority of users are just googling the names or numbers of the PEPs they are looking for based on the number or keywords in the name. It seems the main intended audience of this, and PEP 0 in general, would be the PEP editors, to some extent PEP authors/contributors/sponsors, those who want to actively follow all of Python's development and the SC/PEP Delegates.