Viewing Issue Simple Details
[ Jump to Notes ]
|
[ Issue History ]
[ Print ]
|
ID |
Category |
Severity |
Type |
Date Submitted |
Last Update |
0001754 |
[Issue 8 drafts] Base Definitions and Headers |
Objection |
Error |
2023-06-29 11:15 |
2024-06-11 09:12 |
|
Reporter |
Vincent Lefevre |
View Status |
public |
|
Assigned To |
|
Priority |
normal |
Resolution |
Accepted As Marked |
|
Status |
Closed |
|
Product Version |
Draft 3 |
|
Name |
Vincent Lefevre |
Organization |
Inria |
User Reference |
|
Section |
<float.h> header |
Page Number |
259 |
Line Number |
9089 |
Final Accepted Text |
Note: 0006401 |
|
Summary |
0001754: formula for LDBL_MAX may be incorrect, e.g. for the double-double IBM format |
Description |
For the maximum representable finite floating-point number, the formula (1−b^(−p))b^emax is given, but this formula is incorrect for LDBL_MAX on PowerPC, where the double-double IBM format is used.
For the reference: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61399 [^]
For the LDBL_MAX issue, this was resolved as a defect in the C standard by DR 467, mentioned here: https://www.open-std.org/jtc1/sc22/wg14/www/docs/n2092.htm [^] |
Desired Action |
Do the same change as in the C standard, i.e. add "if that number is normalized, its value is".
Note that the ISO C23 draft has:
"maximum representable finite floating-point number; if that number is normalized, its value is [the formula]" |
Tags |
applied_after_i8d3, issue8 |
|
Attached Files |
|
|