| To: | yyyyyyyyyyyyyyy@xxxxxxxxxxxxx |
|---|---|
| Subject: | Defect in XCU expr |
| From: | yyyyyyy@xxxxxxxxxxx |
| Date: | Fri, 22 Mar 2002 00:16:27 GMT |
Defect report from : Garrett Wollman , FreeBSD Project
(Please direct followup comments direct to yyyyyyyyyyyyyy@xxxxxxxxxxxxx)
@ page 2631 line 16553 section expr comment {GAW-2002b}
Problem:
Defect code : 2. Omission
(Page and line numbers from draft 7 pdfs -- my library is closed
for the day.)
The definition of the expr utility leaves unstated a number of
issues regarding the treatment of arithmetic. This is probably
intentional (so as not to break historic implementations), but it
should be stated explicitly in the text, either in the EXTENDED
DESCRIPTION or in the APPLICATION USAGE, that this is left up to
the implementation. These issues include the range of integer
arguments and the treatment of arithmetic overflow.
Some implementations will detect overflow and return exit status
2 (``invalid expression''); are those implementations conforming?
Likewise division by zero.
Action:
I believe the desire is to avoid constraining historic
implementations. The following text in the EXTENDED DESCRIPTION
would suffice:
The range of /integer/ values in the expr utility is
implementation-defined. If an /integer/ argument is
provided which is outside of this range, or the result
of an arithmetic operation overflows, or a division by
zero is attempted, the output and exit status of the
expr utility are undefined.
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Defect in XCU expr, wollman |
|---|---|
| Next by Date: | Re: Defect in XCU expr, Paul Eggert |
| Previous by Thread: | Re: Defect in XCU expr, Andrew Josey |
| Next by Thread: | Re: Defect in XCU expr, Alexander Terekhov |
| Indexes: | [Date] [Thread] [All Lists] |