Date: Mon, 15 May 1995 15:32:32 -0700 (PDT) From: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com> To: peter@haywire.dialix.com (Peter Wemm) Cc: hackers@FreeBSD.org Subject: Re: Some makefile patches to stop the !#^%!& include files being touched Message-ID: <199505152232.PAA07805@gndrsh.aac.dev.com> In-Reply-To: <Pine.SV4.3.91.950516061736.16341D-100000@haywire.DIALix.COM> from "Peter Wemm" at May 16, 95 06:23:35 am
next in thread | previous in thread | raw e-mail | index | archive | help
> > On Mon, 15 May 1995, Rodney W. Grimes wrote: > > > These were done from a very recent -current.. I got sick of the include > > > files being touched, causing make to rebuild the entire system again. > > > > > > Most of the makefiles were fixed, but these two slipped through: > > > > This is a very small band aid for a much larger problem which I will start > > in on fixing as soon as 2.0.5 is out the door. > > You are not wrong there.. Reinstalling the static libs causes a relink > in /bin and /sbin (among others). I dont mind a relink anywhere as much > as a recompile-the-entire-system because I forgot to reapply the patches > after sup kindly "repaired" them for me. I am not wrong here, there is the huge problem that we have no consitent and correct way to handle the installation of header files for the whole of the source tree (or atleast, not one commited. I have prototype here that does 10 times the job these little hacks of cmp -s do. Any small hack to add more cmp -s's will just get in the way of the proper fix that is coming. I am not talking about rewrite all of .mk, just pulling over my fixes for header file installation (new file bsd.file.mk) that does a pretty damn sweat jobs of handling lots of things, can you say /usr/include is a symlink farm to the source tree with only files symlinked, can you say /usr/include if it is a SHARED=copies type only has files installed into it for the *whole* source tree if the file has changed. Trust me, I've been working on and building *BSD sources for a long time, your little patch is but a drop in the bucket of what needs fixing. > Damn. What a problem! :-) Having recently come from a binary-only > system, the last thing I should be griping about is recompiling the > source... :-) :-) :-), right :-) > Seriously though.. I suspect that a "cmp -s" will fail for many > static libs, merely because of the "ranlib" or "ranlib -t" after the > install.. even though they are otherwise identical. I am not talking about libs, I did not even mention them. I am talking about /usr/include files only. -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation Company Custom computers for FreeBSD
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199505152232.PAA07805>