Date: Fri, 12 Apr 2013 20:09:27 +1000 (EST) From: Bruce Evans <brde@optusnet.com.au> To: Juli Mallett <jmallett@freebsd.org> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, Gleb Smirnoff <glebius@freebsd.org>, src-committers@freebsd.org, Bruce Evans <brde@optusnet.com.au> Subject: Re: svn commit: r249355 - head/lib/libkvm Message-ID: <20130412195828.C1222@besplex.bde.org> In-Reply-To: <CACVs6=9UL9Z8x8BVVM9mLkfF_-=F=%2BUdH-yG_35_-vgH_NuqYA@mail.gmail.com> References: <201304110730.r3B7Uo6d067302@svn.freebsd.org> <20130411175308.Q1435@besplex.bde.org> <20130411082457.GS76816@FreeBSD.org> <20130411184049.W1641@besplex.bde.org> <20130411091418.GW76816@FreeBSD.org> <20130411201652.F1911@besplex.bde.org> <20130411180013.GB76816@FreeBSD.org> <CACVs6=9UL9Z8x8BVVM9mLkfF_-=F=%2BUdH-yG_35_-vgH_NuqYA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 11 Apr 2013, Juli Mallett wrote: > On Thu, Apr 11, 2013 at 11:00 AM, Gleb Smirnoff <glebius@freebsd.org> wrote: >> Bruce, >> >> On Thu, Apr 11, 2013 at 09:07:25PM +1000, Bruce Evans wrote: >> B> Just routine avoidance of namespace pollution. This is easy in such a >> B> simple header. >> >> Sorry, with all respect, but I can't call including sys/types.h >> a namespace pollution. >> >> Ok, even you force me to name it that way, still I would prefer >> namespace pollution instead of handmade copy pasted typedefs. The copying gives a good implementation. Much easier to read (though not write) than definitions in deeply nested includes. We have too many little include files to avoid duplication, but a few big nested include files like machine/_types.h are hard to avoid. > But Gleb, making such changes unilaterally is a bit of a leap. The > project has mostly accepted Bruce's wisdom about trying to minimize > and reduce namespace pollution. Now, this isn't a standard header so > it's quite a bit less of a concern, but it's not no concern. If you > think that we should reverse our trend on including > namespace-polluting headers in system headers, we should discuss that > on arch@, and it shouldn't be something that's done without any > discussion or consideration. > > Should we expect further changes of this nature (and of the proposed > nature removing __size_t and __ssize_t use) if you make changes to > other headers as part of your work? Are you going to add > <sys/types.h> to every header currently using <sys/_types.h> in a > single go, or will you be doing that a little at a time when making > functional changes? That would be a large policy change, not to mention it would break most C99 headers like <stdio.h>. C99 headers can't include POSIX headers like <sys/types.h> since the POSIX namespace rules don't apply to them. Most of them include sys/_types.h or machine/_types.h to get declarararions like __size_t which they use to declare the limited set of typedefs specified by C99. C99 headers are relatively simple and mostly de-polluted. Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20130412195828.C1222>