Date: Wed, 29 Sep 1999 00:44:27 +0800 From: Peter Wemm <peter@netplex.com.au> To: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> Cc: Marcel Moolenaar <marcel@FreeBSD.org>, cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/lib/libc/gen getcwd.c Message-ID: <19990928164427.828B71CBF@overcee.netplex.com.au> In-Reply-To: Your message of "Tue, 28 Sep 1999 12:07:17 -0400." <199909281607.MAA14467@khavrinen.lcs.mit.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
Garrett Wollman wrote: > <<On Tue, 28 Sep 1999 22:44:03 +0800, Peter Wemm <peter@netplex.com.au> said: > > > Hmm.. my experience on other systems suggests sigactions should be > > pre-bzeroed (or memset) for portability reasons. This is required on some > > OS's as there are hidden fields in there, and doesn't require modifying the > > whole tree when a new field is added or changed to struct sigaction. > > Those other systems are broken. POSIX requires that only those fields > defined by the standard need be initialized. Any other fields in > struct sigaction must be ignored unless an implementation-specific > flag is set by the application. (Hence, when POSIX.1b added > sa_sigaction as an alternative to sa_handler, it was necessary to also > add the SA_SIGINFO flag.) I see. That make sense to me, but I don't have a posix spec. It's been a long time (1992 or so) since I regularly did this and it was enough trouble to track down places where people did sa.sa_mask = 0 and that failed if sa_mask was a struct. Fortunately this mostly got hunted down and fixed or we'd be doing it all over again. I withdraw my objections which were based on paranoia and bitter (and outdated) experience. Cheers, -Peter To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19990928164427.828B71CBF>