Text 6091, 99 rader
Skriven 2006-05-18 13:06:04 av Maurice Kinal (1:140/13.1)
   Kommentar till text 6090 av Paul Rogers (1:105/360.0)
Ärende: Export
==============
Hey Paul!

May 17 18:03 06, Paul Rogers wrote to Maurice Kinal:

 PR> Ummm, not hardly!  Just for example, here's my own build script
 PR> for cron/vixiecron:

That's more like a makefile!

 PR> chmod 754 /etc/rc.d/init.d/cron &&
 PR> ln -s ../init.d/cron /etc/rc.d/rc2.d/S22cron &&
 PR> ln -s ../init.d/cron /etc/rc.d/rc2.d/K78cron &&
 PR> ln -s ../init.d/cron /etc/rc.d/rc3.d/S22cron &&
 PR> ln -s ../init.d/cron /etc/rc.d/rc3.d/K78cron &&
 PR> ln -s ../init.d/cron /etc/rc.d/rc4.d/S22cron &&
 PR> ln -s ../init.d/cron /etc/rc.d/rc4.d/K78cron &&
 PR> ln -s ../init.d/cron /etc/rc.d/rc5.d/S22cron &&
 PR> ln -s ../init.d/cron /etc/rc.d/rc5.d/K78cron &&

Unreal.  I've never cared much for that method.  Just start/stop the damn thing
already!!!  :-)

 PR> Not ANY of MY build scripts are that simple!  ;-)

Most of mine are but there are exceptions, util-linux being the worst.  Have
you checked out some of the gentoo build scripts?

 PR> to figure out how to use it.  They really don't need to know
 PR> how to write a fstab.conf in order to do an install--all it
 PR> really needs is to plug in a couple device addresses and the
 PR> script can do that.  And I HATE using vi!

fstab can be quite variable depending on the target, formats and now kernel
version, whether 2.4 or 2.6.  I usually just copy one across and edit it by
hand.  I still find that the simplest and in the case of an initrd it is always
the same so a straight copy is all that is needed there although as in the case
of your cron build script a 'cat' would work such as;

cat > $TARGET/etc/fstab << "EOF"
/dev/ram0               /               ext2    defaults                0 0
none                    /proc           proc    defaults                0 0
none                    /dev/pts        devpts  gid=2,mode=620          0 0
none                    /dev/shm        tempfs  defaults                0 0
EOF

 PR> Naw, a couple sed's can do the job with a template.

Yes but then you still would have to let sed know what partition/device,
format, etc. so somewhere along the line you are inputting this information. 
In the case of a dialog based script it could be checkboxes of all the
available devices/partitions.  

 PR> I'm not
 PR> allowing a lot of variations, but typically an install may
 PR> have different boot, root, & swap devices.

Right.  Mine usually vary on drives such as hda or hdb (<- the usual suspects)
but sometimes sda or sdb or ... sdg.  Heck even hd has gone up to hdf on a few
occassions and in that case there was a special lilo that had to set that to
boot to either hda or sda on the target machine.  In the case of a compactflash
(or usb flash) it was sda on the host and hda on the target and there is no
sensible way for the host machine to know that is going to happen.  It depends
mostly on the target's interface and what device was used on the host machine
to write to whatever device which is seldom, if ever, the same.

 PR> Not during an install--I'm building a new system.  I expect
 PR> I'll be running from a floppy, Tom's RTBT or Trinux, with
 PR> the "source" files on a CD.  That part isn't entirely nailed
 PR> down quite yet.

Right.  I had a simular scheme on a CD once but never ended up using it much. 
I would prefer the usb flash method but unfortunetly that only works on one
machine here at the moment and even there isn't exactly the way I hoped it
would be (stinkin' DOS-think bios).  However that did work to load a 32M
ramdisk yesterday off a compactflash on a usb card reader which was pretty
cool.  Quite a beefy and flexible system but even there I haven't added many
build scripts ... yet.  I left lots of elbow room.

 PR> Not yet.  This is a 2.4.31 system.  It's roughly equivalent to
 PR> a RHL8 or so.  My primary criterion in building it has been
 PR> that it should work effectively on a modest classic Pentium.

I've been using 2.6 kernels with hotplug/udev boots on a 486-33.  It works.

 PR> I use it daily on a 64MB Pentium-MMX 233.  (All but the HX
 PR> chipset don't cache above 64MB.  In that respect the tagline
 PR> is wrong.)  The first thing I did when I decided to build it
 PR> from LFS was write a "mission statement" to guide my decisions.

Yeah.  I need to settle in here and decide what it is I am doing.  However I am
still not happy with the crosscompile thingy but am zeroing in on a few
outstanding issues.  I need to ditch pcmcia on the smp guy though as it is
creating more problems then it is solving lately.

Life is good,
Maurice

--- Msged/LNX 6.2.0
 * Origin: The Pointy Stick Society XIX - Simplify man! (1:140/13.1)