doc: coding_guidelines: replace a whitelist replacement

[mbolivar-nordic]

Nothing in the tree is using the "passlist" term in the "blocklist /
passlist" pair. However, west is using "blocklist / allowlist" already in a
released version.

Recommend blocklist/allowlist instead of blocklist/passlist to avoid
having to add yet another possible replacement, since nothing is using
passlist right now.

Note that blocklist/allowlist is in widespread use throughout many
projects, including Chrome.

Signed-off-by: Martí Bolívar [email protected]

[mbolivar-nordic]

Another reason to prefer allowlist to passlist is that allowlist has the same number of characters as whitelist, so it doesn't affect alignment if you do a big search and replace.

[carlescufi]

Not only is allowlist more understandable, but as @mbolivar-nordic says, we are already exposing our users to this particular term in west.

[pabigot]

Since I started this firestorm in #29498 (review) I'm just going to respond to #29498 (comment) here then move on:

Zephyr doesn't use blocklist.

That was not intended as a statement of policy; it was an insufficiently clear statement of fact: the term blocklist appears in Zephyr mainline only in the west guide documentation and when paired with "passlist". It has not yet been used in Zephyr RTOS code.

You're saying "Zephyr" but I disagree with whoever in the community ruled this out, and the fact that west is already using these terms de facto means they're going to be used by Zephyr users. So I stand firm on keeping them.

That's fine. My position was we had three perfectly adequate names, where the positive and negative verbs are natural antonyms: "allow/deny", "pass/block", and "accept/reject".

I don't find "allow/block" to be a natural pairing. I also am tired of all the bikeshed discussions, and figured that since for this case we actually had a documented set of recommended names it was reasonable to propose sticking with them rather than introducing yet another arguably gratuitous variation.

I don't like changing the list to replace pass/block with allow/block, but if that's how people want to go it's fine. But I don't feel any professional or moral obligation to approve such a change.

[mbolivar-nordic]

I don't find "allow/block" to be a natural pairing. I also am tired of all the bikeshed discussions, and figured that since for this case we actually had a documented set of recommended names it was reasonable to propose sticking with them rather than introducing yet another arguably gratuitous variation.

Given that nobody is using 'passlist', I think that is rather precisely what this PR is doing.