This is part of The Pile, a partial archive of some open source mailing lists and newsgroups.
Subject: Re: rpm and AMD K6 From: jfm2@club-internet.fr Date: Sun, 27 Aug 2000 11:33:59 +0200 (CEST) I built an rpm from source. As I'm building on and for an AMD K6, I said, > "--target=k6." > > I duly created /u02/summer/redhat/RPMSi586-pc-linux-gnu/k6/fetchmail-5.5.1-1.k6 > rpm, but find I can't install it: > > [root@dugite /root]# rpm --upgrade /u02/summer/redhat/RPMSi586-pc-linux-gnu/k6 > /fetchmail-5.5 > 1-1.k6.rpm > package fetchmail-5.5.1-1 is for a different architecture > > unless I specify --ignorearch > > What must I do to avoid having to specify this option each time? > > Unless the package has special assembler instructions like Mesa you shouldn't. In a K6/2 (notice this is not a plain K6) code compiled for K6 is slower that code you let alone ie compiled for 386. === Subject: Re: rpm and AMD K6 From: John Summerfield <summer@OS2.ami.com.au> Date: Mon, 28 Aug 2000 07:27:59 +0800 I built an rpm from source. As I'm building on and for an AMD K6, I said, > > "--target=k6." > > > > I duly created /u02/summer/redhat/RPMSi586-pc-linux-gnu/k6/fetchmail-5.5.1- > 1.k6 > > rpm, but find I can't install it: > > > > [root@dugite /root]# rpm --upgrade /u02/summer/redhat/RPMSi586-pc-linux-gn > u/k6 > > /fetchmail-5.5 > > 1-1.k6.rpm > > package fetchmail-5.5.1-1 is for a different architecture > > > > unless I specify --ignorearch > > > > What must I do to avoid having to specify this option each time? > > > > > > Unless the package has special assembler instructions like Mesa you > shouldn't. In a K6/2 (notice this is not a plain K6) code compiled > for K6 is slower that code you let alone ie compiled for 386. > Jean Are you sure about this? I'm using gcc 2.95 which has special options for the K6 family. Checks... Umm We're both wrong. I have a fetchmail src.rpm to hand; it's trouble-free and doesn't take long to build, so I did this: rpm --rebuild --target k6 /u03/tobuild/fetchmail-5.5.0-1.src.rpm and rpm ran gcc thus: gcc -DHAVE_CONFIG_H -c -I. -I. -O2 socket.c so it's compiling for i386 All I get is an rpm I can't install;-(( Wrote: /u02/summer/redhat/RPMSi686-pc-linux-gnu/k6/fetchmail-5.5.0-1.k6.rpm Now.. It doesn't make sense to me that gcc should produce worse code for the K6 (it IS a K6-II) if I tell it to compile for one than if I don't, but I will compile something and check it out iff 1) I can coax rpm to build properly for the K6 2) I can coax rpm to install for it. Any clues about how I do it? === === Subject: Re: rpm and AMD K6 From: "Mike A. Harris" <mharris@meteng.on.ca> Date: Sun, 27 Aug 2000 10:22:18 -0400 (EDT) On Sun, 27 Aug 2000, John Summerfield wrote: >I built an rpm from source. As I'm building on and for an AMD K6, I said, >"--target=k6." > >I duly created /u02/summer/redhat/RPMSi586-pc-linux-gnu/k6/fetchmail-5.5.1-1.k6 >.rpm, but find I can't install it: > >[root@dugite /root]# rpm --upgrade /u02/summer/redhat/RPMSi586-pc-linux-gnu/k6 >/fetchmail-5.5 >.1-1.k6.rpm >package fetchmail-5.5.1-1 is for a different architecture > >unless I specify --ignorearch > >What must I do to avoid having to specify this option each time? Edit /usr/lib/rpm/rpmrc... it is more or less self explanatory. This begs a question... Why is rpmrc in /usr/lib/rpm/? Should it not make more sense in /var/lib/rpm/? I mean this is a configuration file, and it is in /usr which is sometimes mounted read-only. Not a very good idea IMHO. It would make _much_ more sense under /var/... If your /usr is read only, move the file to /var/lib/rpm and make a symlink in /usr/lib/rpm.. TTYL === Subject: Re: rpm and AMD K6 From: Matt Wilson <msw@redhat.com> Date: Sun, 27 Aug 2000 19:31:47 -0400 NO NO NO! /etc/rpm/macros and /etc/rpm/rpmrc are for local changes! === Subject: Re: rpm and AMD K6 From: Chris Abbey <cabbey@bresnanlink.net> Date: Sun, 27 Aug 2000 20:55:20 -0500 At 19:31 8/27/00 -0400, Matt wrote: >NO NO NO! /etc/rpm/macros and /etc/rpm/rpmrc are for local changes! well, I don't know about Mike, Jean or John... but I for one am not psychic... perhaps a nice little message at the top of /usr/lib/rpm/ copies of the files along the lines of: # This is a static RPM configuration file, local user edits should # be done in the /etc/rpm/ files not here. would be appropriate? -=Chris === Subject: Re: rpm and AMD K6 From: Matt Wilson <msw@redhat.com> Date: Mon, 28 Aug 2000 03:18:11 -0400 On Mon, Aug 28, 2000 at 01:09:25AM -0400, Mike A. Harris wrote: > > Sorry... I couldn't find that in the nonexistant RPM manual. > ;o) Actually.. it probably is in Maximum RPM even though it's > outdated, so I'll shut up now. ;o) Maximum RPM, index: "rpmrc, locations of: 368" On page 368: Next, we have /etc/rpmrc. B 2.2 /etc/rpmrc The file /etc/rpmrc, unline /usr/lib/rpmrc, is fair game for modifications and additions. In face, /etc/rpmrc isn't created by default, so its contents are entirely up to you. It's the perfect place to keep rpmrc entries of a system-wide or global nature. ... Cheers, === Subject: Re: rpm and AMD K6 From: jfm2@club-internet.fr Date: Mon, 28 Aug 2000 22:29:14 +0200 (CEST) Unless the package has special assembler instructions like Mesa you > > shouldn't. In a K6/2 (notice this is not a plain K6) code compiled > > for K6 is slower that code you let alone ie compiled for 386. > > > > > Jean > Are you sure about this? I'm using gcc 2.95 which has special options for the > K6 family. > In fact what I told was valid for gcc 2.96. For gcc 2.95 at O2 optimization level and using the Byte benchmarks the code generated with arch=k6 (optimize for k6, compiler is free to pick k6 specific instructions) is: 1% slower for memory-intensive tests, 1.5% faster for integer tests, 2% slower at floating point tests respective to code targetted at 386. Tests were run on a K6/2 450. In othzer words either gcc 2.95 does not amke a good job at optimizing specifically for K6 or the K6/2 is internally different enough from the K6 to make K6 optimizations ineffective. === Subject: Re: rpm and AMD K6 From: John Summerfield <summer@OS2.ami.com.au> Date: Tue, 29 Aug 2000 07:53:13 +0800 On Mon, Aug 28, 2000 at 01:09:25AM -0400, Mike A. Harris wrote: > > > > Sorry... I couldn't find that in the nonexistant RPM manual. > > ;o) Actually.. it probably is in Maximum RPM even though it's > > outdated, so I'll shut up now. ;o) > > Maximum RPM, index: "rpmrc, locations of: 368" > > On page 368: > > Next, we have /etc/rpmrc. > > B 2.2 /etc/rpmrc > > The file /etc/rpmrc, unline /usr/lib/rpmrc, is fair game for > modifications and additions. In face, /etc/rpmrc isn't created by > default, so its contents are entirely up to you. It's the perfect > place to keep rpmrc entries of a system-wide or global nature. > well, yes, but M RPM applies to rpm 2.5, doesn't it? rpm 3.0 rearranged things to GKW, and knowing some of the information in that book is obsolete I don't know what to trust. I'm not psychic either. === Subject: Re: rpm and AMD K6 From: John Summerfield <summer@OS2.ami.com.au> Date: Tue, 29 Aug 2000 07:47:20 +0800 Edit /usr/lib/rpm/rpmrc... it is more or less self explanatory. That's error-prone; it gets replaced whenever I upgrade rpm, and I seem to be forced to do that altogether TOO often lately. > > This begs a question... > > Why is rpmrc in /usr/lib/rpm/? Should it not make more sense in > /var/lib/rpm/? I mean this is a configuration file, and it is in > /usr which is sometimes mounted read-only. Not a very good idea > IMHO. It would make _much_ more sense under /var/... ro if fine for that configuration file IFF users don't ever need to change it. User-changeable configuration files surely belong in /etc The rpm database belongs under /var (it's "state" information). === Subject: Re: rpm and AMD K6 From: "Mike A. Harris" <mharris@meteng.on.ca> Date: Tue, 29 Aug 2000 00:50:16 -0400 (EDT) On Tue, 29 Aug 2000, John Summerfield wrote: >Date: Tue, 29 Aug 2000 07:47:20 +0800 >From: John Summerfield <summer@OS2.ami.com.au> >To: redhat-devel-list@redhat.com >Content-Type: text/plain; charset=us-ascii >Subject: Re: rpm and AMD K6 > >> >> Edit /usr/lib/rpm/rpmrc... it is more or less self explanatory. > >That's error-prone; it gets replaced whenever I upgrade rpm, and I seem to be >forced to do that altogether TOO often lately. When I edit files like that, I then do a "chattr +i filename" and it solves that problem. ;o) Perhaps a bad thing to do, but if it breaks, I get to keep both pieces. ;o) === Subject: Re: rpm and AMD K6 From: jfm2@club-internet.fr Date: Tue, 29 Aug 2000 08:59:19 +0200 (CEST) On Tue, 29 Aug 2000, John Summerfield wrote: > >> Edit /usr/lib/rpm/rpmrc... it is more or less self explanatory. > > > >That's error-prone; it gets replaced whenever I upgrade rpm, and I seem to be > >forced to do that altogether TOO often lately. > > When I edit files like that, I then do a "chattr +i filename" and > it solves that problem. ;o) Perhaps a bad thing to do, but if > it breaks, I get to keep both pieces. ;o) > > Emacs automatically saves old files. It can even put them under version control if it sees a CVS or RCS directory. This way you don't lose the original file in case you save repeatedly. And it makes a nicer info reader. ;-) === Subject: Re: rpm and AMD K6 From: John Summerfield <summer@OS2.ami.com.au> Date: Tue, 29 Aug 2000 15:23:33 +0800 Edit /usr/lib/rpm/rpmrc... it is more or less self explanatory. > > > > > >That's error-prone; it gets replaced whenever I upgrade rpm, and I seem to > be > > >forced to do that altogether TOO often lately. > > > > When I edit files like that, I then do a "chattr +i filename" and > > it solves that problem. ;o) Perhaps a bad thing to do, but if > > it breaks, I get to keep both pieces. ;o) > > > > > > Emacs automatically saves old files. It can even put them under > version control if it sees a CVS or RCS directory. This way you don't > lose the original file in case you save repeatedly. And it makes a Of course, neither of these techniques solves the problem of rpm clobbering the most-desired version of the file However, as rpm apparently reads some files from /etc (if they exist), that seems a sensible way to go. It's easy to undo too;-) ===