[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Parted 1.5.1-pre

From: Andrew Clausen
Subject: Re: Parted 1.5.1-pre
Date: Fri, 02 Feb 2001 18:09:42 -0200

Dan Knapp wrote:
>   The more templatized specifications may or may not be.  Distribution
> installers can always write a sequence of commands to be executed by the
> partitioning software of their choice, and are much more likely to be able
> to capture the needs of a variety of systems in an automated way than a
> simple file with parameters such as "fill rest of disk" and "allocate 128M
> from the end as swap" is.  There's certainly a use for "flat mirroring" -
> for example, when you buy 20 or 200 machines at the same time, you could
> set up the partition table then transfer the filesystems over the network.
> Of course, just because they're sold as the same model doesn't mean all their
> components are identical...  Are installers really that bad that it's
> preferable to forgo them, though?

I don't think so.  I think installers are good, and getting better ;-)
You can set up dhcp + netboot (right name?) + nfs, and automagically
install an entire network of machines...

I know we can do this with MI, anyway, and I think kick start (red hat)
can to.

>   However, it certainly would be possible to find some semantics that would
> allow one to omit the details that are going to be changed between
> restorations (on the principle of minimizing the number of parsers in a given
> program, sure, no reason not to make the backups and the configuration files
> use the same format) and specify how they are to be calculated.

I'm not sure I like the idea of requiring libparted to link against
libxml... (or whatever).  It's too big already for many rescue/boot

I don't think backups need to be human-editable, although human
readable might be nice.  I think something like

fprintf (stream,
         "id-%d: start-%l end-%l c-%d h-%d s-%d type-%d\n",
         num, start, end, c, h, s, type);

would suffice.

I don't think other programs would/should ever need to parse these.

> > I think having configuration files to store such setups is a good
> > idea, but it's different to backing up partition tables (which is
> > also useful)
>   I would even say that those configuration files would be redundant.
> Scripting suffices and is probably more versatile.

I think configuration files (which could be generated by GUIs) could
be a lot easier.  Which isn't to say we shouldn't allow scripting.

Andrew Clausen

reply via email to

[Prev in Thread] Current Thread [Next in Thread]