home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.arch
- Path: sparky!uunet!munnari.oz.au!uniwa!cujo!cc.curtin.edu.au!zrepachol
- From: zrepachol@cc.curtin.edu.au
- Subject: Re: Any use for Branch if Even/Odd ?
- Message-ID: <1992Dec28.211934.1@cc.curtin.edu.au>
- Lines: 15
- Sender: news@cujo.curtin.edu.au (News Manager)
- Organization: Curtin University of Technology
- References: <endecotp.723992157@cs.man.ac.uk> <1g8ub1INN8it@agate.berkeley.edu> <GJR.92Dec11230751@chamarti.ai.mit.edu>
- Date: Mon, 28 Dec 1992 12:19:34 GMT
-
- In article <GJR.92Dec11230751@chamarti.ai.mit.edu>, gjr@zurich.ai.mit.edu (Guillermo J. Rozas) writes:
- ...
- > The alpha has those too, but that is not the issue. If I remember
- > correctly, all odd integer values were considered true in BLISS, and
- > all even values false. Using test-for-zero for boolean tests only
- > works if zero is the unique representation for one of the boolean
- > values (as in C), but this is neither inherent nor clearly superior.
-
- Close, but you have fallen into the hole that the BLISS spec trys to avoid.
- The value of the fullword is irrelevent, only the low bit counts. Its value as
- an int or float or char is irrelevent to the true/false test. Note that the
- odd/even true/false mapping fails with floats or scaled quantities.
-
- ~Paul
-
-