Review requested:

• @nodejs/crypto
• @nodejs/gyp
• @nodejs/http
• @nodejs/net
• @nodejs/tsc
• @nodejs/v8-update

CI: https://ci.nodejs.org/job/node-test-pull-request/43290/

Similar: #35051

I may be behind the times on what abbreviations we can expect people to be familiar with, but I think this is going to confuse people more than enlighten them. Also FWIW the style guide doesn't seem to recommend this. https://docs.microsoft.com/en-us/style-guide/a-z-word-list-term-collections/term-collections/bits-bytes-terms

I think this is going to confuse people more than enlighten them

That's likely true. For memory sizes, for example, GB almost exclusively means base 2 anyway, whereas things are vastly different for data transmissions etc.

I agree.
I'm not even sure that in all changed places those are actually, for instance KiB vs KB. If we want to advance this PR, we should double check just in case.

Anyway, this PR LGTM if we want to use those units.

I'm further confused by people both approving this PR and agreeing that it is likely to increase confusion.
¯\(ツ)/¯

Lol. You are definitely right.

I meant to say that the code changes look fine and we could definitely land this to improve docs(up for me on this).

But, at the same time, the possible confusion is definitely worth a second evaluation (in contrast to handle this like a trivial doc fix PR).

Hope this rant was understandable :)

My take on this is we should use the technically correct units in our code and comments, I think it's fair to assume that most folks who would read it would not be confused by them.
For our docs, I understand that confusing the reader is not something we want. Maybe a good mitigation would be to always specify the raw value next to the abbreviated unit (e.g. 16 KiB (16384 bytes)).
@Trott wdyt?

The standard that defined KiB as the abbreviation for 1024 bytes also says that the unit is a kibibyte and not a kilobyte (which is 1000 bytes, according to the standard). I don't think any of this is at all widely adopted and I'm not sure I want Node.js docs to be an early adopter for this type of thing, especially when "early adopter" means doing it after 15 years of no significant adoption of the terms.

This from 2009 still seems to be the case: [Why does Explorer use the term KB instead of KiB?}(https://devblogs.microsoft.com/oldnewthing/20090611-00/?p=17933):

If you look around you, you’ll find that nobody (to within experimental error) uses the terms kibibyte and KiB. When you buy computer memory, the amount is specified in megabytes and gigabytes, not mebibytes and gibibytes. The storage capacity printed on your blank CD is indicated in megabytes. Every document on the Internet (to within experimental error) which talks about memory and storage uses the terms kilobyte/KB, megabyte/MB, gigabyte/GB, etc. You have to go out of your way to find people who use the terms kibibyte/KiB, mebibyte/MiB, gibibyte/GiB, etc.

For our docs, I understand that confusing the reader is not something we want. Maybe a good mitigation would be to always specify the raw value next to the abbreviated unit (e.g. 16 KiB (16384 bytes)).
@Trott wdyt?

I don't love it, but it does have the virtue of precision.

Another option is to always use bytes: 16384 bytes. That will work up to a point for relatively small values, but not so much once we're talking about gigabytes or gibibytes.

(All my cranky comments are non-blocking. I just want these things considered before we land.)

I'm all for favoring correctness where it matters, and especially in scientific writing, I try to consistently use IEC symbols (KiB, MiB, etc.). That being said, according to the reference @aduh95 provided, using the unit symbols KB, MB, etc. for base-2 units is not incorrect but merely legacy:

Another possibility (that might not always be practical) would be to use the common/"legacy" abbreviations but precise byte counts for clarity: 16 KB (16384 bytes)

I like this one.

Given that this PR has enough approvals + wait time, I'm down to land it as it, and to improve on follow up PRs. FWIW I'd be disappointed if we had to settle on using the legacy units, but as long as we remove the confusion of when we're using metric and when we're using base-2, I'm fine with it.

Landed in 1e76165