thanks for your report: since i don't have intel macs and our intel testers haven't recent cpp ones i could not be 100% sure that universal installations works as expected.

:-)

PETER DA SILVA said:

> Looks like 98% of the code is there to >support the italian translation, or is >commented out.

well, it is something less than 98%… ;-)

anyway,here is my explanation:
1) i started the developing in an italian mac community and i'm italian myself, so lots of comments are in that language…
now that the project is international i'm making only english comments and traslating (or, if obsolete, removing) the old ones.

2)there is a big block of code commented (starting at line 65 of installer.sh, if you want take a look at it…). this code as a story: it should implement a feature that will be added in a future version, a way to install a very minimal system (about 500mb on ppc, 1gb on intels for now). it was added to our -unstable branch, but it caused a lot of trouble (see bug 1651505 here: https://sourceforge.net/tracker/index.php?func=detail&aid=1651505&group_id=186964&atid=919399 ), so it was commented and postponed. for simplicity reasons the tree has been synced, so this commented code is now also in -testing (the tree which we are using for these releases). it is anyway armless, so there was no reason to make thing complicated keeping the trees unsynchronized…

>It should work for a completely vanilla
>install, but it's not really robust.
what do you mean?

>If you want to set up a boot disk for a
>limited version of OS X, look at
>XPostFacto, and set up a recovery disk
>which creates a RAM-disk install for
>just the applications you need. It's
>PPC-only, but so is this.

well, the aims of the xpostfacto team is very different from the ours, though i admit that there can be some overlays…

>If you want to set up a boot disk based
>on your existing system, look at Carbon
>Copy Cloner.
We're not making disks based on previously installed systems. i use ccc for my backups and i'm happy with it, but it does totally different things.

Well, best of luck. You probably ought to do a GUI of some kind (Applescript or Tcl/Tk would be easiest). The shell code is a little eccentric, like, using "ls" rather than "test -d" to see if a directory exists... which is the main reason I'm concerned about the fragility of the code.

a gui (applescript) is in the -unstable branch of the svn, it will be part of the next version…

>The shell code is a little eccentric,
>like, using "ls" rather than "test -d" to
>see if a directory exists... which is the
>main reason I'm concerned about the
>fragility of the code.

well, i'm completely self-taught programmer, i was sure that there were better methods, but i had no idea of how to do some things in a "usual" way. thanks for the hint, i will use test -d in the future! (2/11/2007, Version: 0.0.1b1)

There's a lot of capability in the shell.. try "man sh", "man test", and so on...

>There's a lot of capability in the
>hell.. try "man sh", "man test", and so
>on...
I know that… it is not so immediate to imagine what is the best way to do something…

The thing that really helped me the most was looking at how complex and portable scripts other people wrote worked. Try figuring out the configuration script of any substantial software package on Sourceforge. Once your brain quits exploding you'll be 10 times the shell hacker. :)

The thing that really helped me the most was looking at how complex and portable scripts other people wrote worked. Try figuring out the configuration script of any substantial software package on Sourceforge. Once your brain quits exploding you'll be 10 times the shell hacker. :)

I agree. ;-)
E.G. the darwin installer was really useful…