[Cfp-interest] terms DPD and BID

[Jim Thomas]

At the Feb CFP meeting, Mike suggested we use IEC 60559 terminology, instead of Densely Packed Decimal (DPD) and Binary Integer Decimal (BID), for the two encoding schemes for decimal formats. Mike made the point that DPD generally refers to the encoding of decimal digits (rather than to a format encoding as in our current draft).

Note, on the other hand, that Binary Integer Decimal (BID) is used in the literature to refer to the format encoding, consistent with the way it's used in our current draft.

Would it suffice to add footnotes something like: 

"Densely Packed Decimal (DPD) is generally used to refer to a particular binary encoding scheme for decimal digits. In this document it used to refer to the IEC 60559 encoding for decimal formats based on DPD encoding of the significand. IEC 60559 refers to DPD encoding (of the significand) as decimal encoding."

and

"BID is, in IEC 60559 terminology, the decimal format encoding based on binary encoding of the significand.

The difficulty we had, and still have, in following IEC 60559 terminology, is that IEC 60559 avoids introducing terms to designate the format encodings per se. In 3.5.2 it says:

"The encoding within these fields [combination field G and trailing significand field T] depends on whether the implementation uses the decimal or the binary encoding for the significand.

"1)  If the implementation uses the decimal encoding for the significand, …

"2)  Alternatively, if the implementation uses the binary encoding for the significand, …"

Thus it defines decimal and binary encodings of just the significand. 

If the footnotes approach above isn't satisfactory, we might say in (Part 2) 12.3 something like:

"The types
…
represent values of decimal floating types in one of the two alternative encodings allowed for decimal formats by the IEC 60559 standard: the encoding based on decimal encoding of the significand or the encoding based on binary encoding of the significand, "

and in 12.4.2:

"The encodedpd functions convert the argument into the encoding based on decimal encoding of the significand. …"

and similarly for the other re-encoding functions

and add something about our use of the acronyms dpd and bid in type and function names, or else change the names.


-Jim








-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20130219/3fe33ae3/attachment.html