Attached is the diff between my working copy and -06. Comments and suggested text welcome. I wonder if we need to add a definition of "leap second table expiration". On 3/7/23 5:59 PM, Paul Eggert wrote:
In rereading the draft I found a couple of technical issues.
1. Most of this draft was written before the recent announcement that leap seconds are possibly being discontinued. As I understand it, this proposal is likely but not yet set in stone. Perhaps we need a sentence or two about this in the draft? Something like the following at the first mention of leap second expiry:
"If leap seconds become permanently discontinued, as requested by the General Conference on Weights and Measures[*], the leap second table SHOULD NOT expire since it will not be updated in the foreseeable future."
and cite Resolution 4 of the 27th CGPM (2022) <https://www.bipm.org/en/cgpm-2022/resolution-4>.
2. The use of TZ strings like 'XXX3EDT4,0/0,J365/23' is described confusingly, since 3.3.1 says that this is not an extension to POSIX (i.e., we're just spelling out what POSIX says, so this is not a TZ string extension), whereas the title of 3.3.1 says "TZ String Extensions". Because section 4 says a writer should generate a version 3 file (which can contain TZ string extensions) only if necessary, it's too easy to read this as saying that if the TZif file contains a TZ string like 'XXX3EDT4,0/0,J365/23' then it must be at least version 3 - which is not how current zic behaves (it's just version 2). (This confusion is my fault; sorry.)
To fix this, how about if we move the text and example in the bullet point "* DST is considered to be in effect all year ..." up from section 3.3.1 to section 3.3, with minor wordsmithing to continue to make it clear that we're only spelling out POSIX more explicitly, not extending it. As a result of this wordsmithing, there will only be one POSIX extension (extending hours to -167 through 167).
I suppose we should also should note that a future version of POSIX is likely to adopt the POSIX extension documented in 3.3.1. Even if that happens, 3.3.1 will still considered to be an extension, in that TZif files using it will still need to be marked version 3 or later.
-- Kenneth Murchison Cyrus Development Team FastMail Pty Ltd