[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

"reasonability" check



Given the replies to Chuck and my notes (COBOL and what will be required 
when/if it tries to be fully conforming), let me see if I understand what has 
been said - and what we MIGHT do:

1) the FormatOf functions provide conversion support from one format to another 
(these may both be 754-defined floating point items or one may be such and the 
other one of a list of required numeric formats (e.g. decimal strings, hex 
strings, etc).

2) The COPY (x) function provides transfer of data between two identically 
defined items.

3) The above two are mutually exclusive:  Copy doesn't define results for 
differently defined items and FormatOf doesn't define processing between 
identically defined formats.

4) 754 requires certain "functions" to be provided by a "conforming" 
programming language. As long as those functionalities are provided, ADDITION 
(even similar) secondary functions may work the same - but they may also work 
differently than the comparable 754 feature.

   * * * *

If the above is true, then COBOL which has two separate features COPY and 
COMPUTE *could* do one of two things and be 754-conformign:

A) implement COMPUTE for FormatOf of differently defined items (and how it 
works for identically defined items is up to COBOL) WHILE using MOVE for copy 
of identically defined items  and how it works for differently defined items is 
up to COBOL.

  or

B) Use  COMPUTE to implement FormatOf for differently defined items and to also 
implement COPY for identically defined items - while how MOVE works for both 
identically and differently defined items would be up to COBOL.

NOTE: The COBOL "NumVal-????" built-in functions might be needed and might need 
enhancement to handle conversion from certain types of "strings".

   * * * *

Does this analysis pass a "reasonability" test from non-COBOL IEEE 754 experts?


754 | revision | FAQ | references | list archive