Date: Thu, 11 Jan 2001 00:36:17 -0500 (EST) From: Daniel Eischen <eischen@vigrid.com> To: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> Cc: cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: errno in libc (was Re: cvs commit: src/sys/alpha/alpha interrupt.c machdep.c mp_mac) Message-ID: <Pine.SUN.3.91.1010111001858.29952A-100000@pcnet1.pcnet.com> In-Reply-To: <200101110515.AAA42400@khavrinen.lcs.mit.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 11 Jan 2001, Garrett Wollman wrote: > <<On Thu, 11 Jan 2001 00:12:07 -0500 (EST), Daniel Eischen <eischen@vigrid.com> said: > > > Should I discard these changes, or is _set_errno(x) seen as a better > > interface than errno = x? > > The latter is required to work in the API. Library internals may be > different, but I would suggest that breaking `errno' just in the > library would be a poor choice. I'm not changing any interfaces or breaking anything. I just changed all libc internal settings of errno to use _set_errno(x) instead of errno = x. Using errno = x in libc will still work. I guess the question is, what would we rather see used internally in libc? It should probably be consistent either way. I've already done the grunt work of changing errno = x to _set_errno(x), so the effort to convert shouldn't be given any weight. I also don't care if I have to discard my changes. I'd rather see _set_errno(x) used internally in libc; it just seems easier to see that it can handle the threaded case. Using errno = x faked me out so much that I changed it ;-) But whatever -- I have no strong feelings either way. -- Dan Eischen 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?Pine.SUN.3.91.1010111001858.29952A-100000>