gh-64612: Update error handlers list under open()

[StanFromIreland]

• Issue: Errors in documentation of standard codec error handlers #64612

---

📚 Documentation preview 📚: https://cpython-previews--137304.org.readthedocs.build/

[encukou]

I'm not convinced the two-column table, with name in the first column and prose in the second, is better than a buletted list. (Especially as two tables with unaligned columns.)

The "reproduced below for convenience" sounds like the tables should be the same. Perhaps "summarized below for convenience" would be better, with additional details left out?

[encukou]

Aren't these all Unicode codecs?

Suggested change

	- Only available for Unicode codecs.
	- Only available for UTF-8, UTF-16 and UTF-32 codecs.

[StanFromIreland]

The codecs documentation lists the little/big endian variants, though I think wr can be less specific here.

[encukou]

We can, but “Unicode codecs” sounds like a proper term, while I see no definition that would link it to the UTF-{8,16,32} codecs specifically.

[StanFromIreland]

I'm not convinced the two-column table, with name in the first column and prose in the second, is better than a buletted list

I see, it is a table in the codecs docs and so, in an attempt to make them consistent converted it to (albeit a more convenient form of) a table.

[encukou]

With 3-5 lines per entry, it could just as well be a copy of the original table.
What about making the summary super brief, something like:

• strict: raise UnicodeError
• ignore: omit malformed data
• replace: replace with ? or �
• backslashreplace: replace with \xhh, \uhhhh, or \Uhhhhhhhh

and so on?

[encukou]

We can, but “Unicode codecs” sounds like a proper term, while I see no definition that would link it to the UTF-{8,16,32} codecs specifically.

[StanFromIreland]

Counter proposal, why bother with super short summaries when we can just link straight to the main table?

[encukou]

That sounds good, too!