This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
Although we have taken IEEE754r as a prime motivation for doing precisionDecimal, we seem not to follow the precedent for other types, in that we do not say in so many words: "this type is intended, with a few exceptions described below, to represent the values of IEEE754r floating point decimal", or whatever the right way to say this is. The exceptions include the fact that we have only a single NaN, and maybe some others I'm forgetting. Furthermore, it might be useful to state non-normatively, in an appendix or otherwise, what the correspondences are. I.e. to explain how to map value space properties such as artithmeticPrecision to their equivalents in IEEE754r.
Discussed at 2005-11-18 telecon. RESOLUTION: Phase 1 agreement: Editors to work with Noah to draft such a Note, and identify a location for it, bring back to the WG.
Dave Peterson sent a proposal to the WG on 18 November, proposing that the sentence The special values are introduced to make the datatype correspond closely to the floating-point decimal datatypes described by the forthcoming revision of IEEE/ANSI 754. be replaced by This datatype is introduced to make a variant of decimal that corresponds more closely to the floating-point decimal datatypes described by an expected forthcoming revision of IEEE/ANSI 754. "Precision" of values is retained, and the special values (two zeros, infinities, and not-a-number) are added to that end. Unless I'm mistaken, this is now ready for WG action. I've revised the summary statement to make it more revealing when truncated by Bugzilla's list view.
Two amended versions have been proposed in email recently: (A) speaks of "the" revision of 754, not "a" revision, and moves "to that end". Diffed against the earlier proposal, it is: This datatype is introduced to make a variant of decimal that corresponds more closely to the floating-point decimal datatypes described by [an]^the^ expected forthcoming revision of IEEE/ANSI 754. ^To that end,^ "Precision" of values is retained, and the special values (two zeros, infinities, and not-a-number) are added [to that end]. (B) follows (A) but deletes "to that end" entirely: This datatype is introduced to make a variant of decimal that corresponds more closely to the floating-point decimal datatypes described by [an]^the^ expected forthcoming revision of IEEE/ANSI 754. "Precision" of values is retained, and the special values (two zeros, infinities, and not-a-number) are added [to that end].
After further discussion, the editors have converged on the following wording (similar to B, but with 'provide' instead of 'make'): This datatype is introduced to provide a variant of decimal that corresponds more closely to the floating-point decimal datatypes described by [an]^the^ expected forthcoming revision of IEEE/ANSI 754. "Precision" of values is retained, and the special values (two zeros, infinities, and not-a-number) are added [to that end].
The ^ (add) and [] (deletion) markings appear to cause some confusion. The clear-text form of the editors' consensus proposal is: This datatype is introduced to provide a variant of decimal that corresponds more closely to the floating-point decimal datatypes described by the expected forthcoming revision of IEEE/ANSI 754. "Precision" of values is retained, and the special values (two zeros, infinities, and not-a-number) are added.
I still am a bit confused about the phrase "more closely", as we don't say more closely than what. I'd be tempted to leave out the word "more", leaving: "This datatype is introduced to provide a variant of decimal that corresponds more closely to the floating-point decimal datatypes described by [an]^the^ expected forthcoming revision of IEEE/ANSI 754." Not worth a lot of discussion if we don't get pretty quick consensus. Thanks! Noah
What I meant to say was: I still am a bit confused about the phrase "more closely", as we don't say more closely than what. I'd be tempted to leave out the word "more" (and moving the word "closely"), leaving: "This datatype is introduced to provide a variant of decimal that closely corresponds to the floating-point decimal datatypes described by [an]^the^ expected forthcoming revision of IEEE/ANSI 754." Not worth a lot of discussion if we don't get pretty quick consensus. Thanks! Noah
At its telcon of 13 January, the WG adopted the proposal in comment 5 as amended by comment 7. The change was integrated into the status quo documents of 13 January.