[Cfp-interest 2997] Annex G

[Jim Thomas]

Following up on the CFP discussion of N3206 “The future of imaginary types”,  I think there are some changes that could be considered.

1. The definitions of the +, -, and * operators with a real and a complex operand and of the / operator with a complex numerator and a real denominator could be moved from Annex G to the main body of the standard. This would nail down the semantic of these operations, in a way that was consistent with IEC 6009 and that was efficient to implement.

2. Along with change 1, the definition of imaginary types and the definitions of the  +, -, and * operators with an imaginary and a complex operand and of the / operator with a complex numerator and an imaginary denominator, and other rules for imaginary that do not depend on IEC 60559 could be moved from Annex G to the main body of the standard. Imaginary types could still be optional. The purpose of this change would be to improve the locality of the various parts of the specification for imaginary types.

3. Annex G could treat imaginary types as an optional feature rather than a requirement of the annex. I think this would be better if done together with changes 1 and 2, but could be done without them by just making imaginary an option feature of Annex G. The purpose of this change would be to encourage IEC 60559 implementations to support IEC 60559 related specification in Annex G, even it the implementation does not provide imaginary types.

Although CFP’s stated purpose has been C binding for IEC 60559, I believe we could address this matter which is strongly related to IEC 60559 compatibility. CFP has addressed other issues that are marginal to its purpose, e.g. accommodation for double-double.

I suggest CFP consider investigating the options and drafting a C26 proposal for WG14. 


- Jim Thomas




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20240206/4583456b/attachment.htm>