Date: 16 Apr 1999 09:09:09 -0400 From: Lowell Gilbert <lowell@world.std.com> To: freebsd-current@freebsd.org Subject: Re: swap on Irix (overcommiting, etc.) Message-ID: <rd6btgozp8q.fsf@world.std.com> In-Reply-To: Mikhail Teterin's message of Thu, 15 Apr 1999 15:34:54 -0400 (EDT) References: <199904151934.PAA98792@misha.cisco.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Mikhail Teterin <mi@misha.cisco.com> writes: > Sorry. I'm just repeating what Ladavac Marino wrote in > <55586E7391ACD211B9730000C11002761795EB@r-lmh-wi-100.corpnet.at>: > > LM: Please note that memory overcommit architectures are a > LM: rather common optimization; FreeBSD is one of them. They > LM: do, however, break the ISO/ANSI C conformance (strictly > LM: speaking). > > Since there was no immediate (nor later) rebuttal, I assumed, that > everyone quietly agreed... Absolutely not. It is *definitely* okay for system conditions outside the realm of the C spec to effect the execution of "conforming" programs. Otherwise, having a system shutdown interrupt the program would be enough to make the system non-conforming. Heck, the *existence* of kill(1) and SIGKILL would be enough to make for a non-conforming C environment. The system running short on virtual memory (whether it be by having a user program touch memory it had previously allocated, or by having a new user log in, or a new sendmail daemon starting up) fits squarely into that category. How (and when) you assign backing store to virtual memory is a *very* interesting topic, but the ISO C compliance issue is a red herring, even to the most pointlessly pedantic language lawyers. Be well. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?rd6btgozp8q.fsf>