Date: Sun, 15 Feb 2009 01:01:49 +0100 From: Luigi Rizzo <rizzo@iet.unipi.it> To: Bruce Evans <brde@optusnet.com.au> Cc: svn-src-head@FreeBSD.org, Luigi Rizzo <luigi@FreeBSD.org>, src-committers@FreeBSD.org, svn-src-all@FreeBSD.org Subject: Re: svn commit: r188578 - head/sys/netinet Message-ID: <20090215000149.GB95636@onelab2.iet.unipi.it> In-Reply-To: <20090215052941.E2522@delplex.bde.org> References: <200902131514.n1DFEhft091837@svn.freebsd.org> <20090214183758.I847@besplex.bde.org> <20090214175136.GB89369@onelab2.iet.unipi.it> <20090215052941.E2522@delplex.bde.org>
index | next in thread | previous in thread | raw e-mail
On Sun, Feb 15, 2009 at 06:39:56AM +1100, Bruce Evans wrote: ... > BTW, I also disagree with you adding comments to include lines to say > what the includes are for. There is no chance that such comments are > complete or can be maintained. For completeness, every include line > would need a paragraph or two like some of the ones in this mail to > explain all the reasons why a header is included, and the details will > change when code is added. The reasons why a header is needed are > most easily seen by trying to compile without it. A generic header > like <sys/types.h> will be needed for a myriad reasons and listing > them all is not useful. A small header will be needed for only a small > number of reasons and listing them all is even less useful, since it > is easy to find and understand the reasons if only a couple of > declarations are used. All true. But I am only annotating "non-obvious" or questionable #include, and only for the details relevant in the context of that source file. E.g. if I am including <net/if.h> just to know what IFNAMSIZ is, or <net/route.h> just to get a forward declaration for 'struct route', that's worth mentioning in my opinion, because those #includes are possible candidates for removal once the required information is obtained elsewhere (or it is not needed anymore). cheers luigihelp
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090215000149.GB95636>