Re: GPT as default?

On Apr 22, 2007, at 12:22 PM, Ivan Voras wrote:

> Marcel Moolenaar wrote:
>
>> These are my requirements (FWIW):
>> 1. It needs to have a friendly UI -> I'm using a curses based  
>> interface.
>
> Regarding this part: one of the possible side-effects of the installer
> might be a GUI for disk partitioning. It's not a text-mode UI, but  
> maybe
> we can avoid duplication of effort here :)

I'm all ears. Even if we can't avoid some duplicate effort in the UI
area, we should be able to share the backend logic.  Also, I'm
interested to learn about how you figured the UI to look like.  It
may be good, in case we can't share the UI, to have it look roughly
the same?

>> 2. It needs to be used in scripts -> It needs a similar usage  
>> model as
>>    gpt(8).
>> 3. It needs to replace gpt(8) -> see point 2.
>
>> Points 2 and 3 are important for our release process.
> Yes. As far as I can tell, they are also enough for today - the  
> rest are
> useful, but not necessary "bells and whistles".

I tend to agree. However, gpt(8) is the tool that works for today
on ia64 and there's not even a tool for PowerPC. I don't feel I'm
advancing anything if I replace gpt(8) with a tool that merely works
for today. The g_part geom is designed to unify all partitioning
schemes and I think that a unified tool is very appropriate. This
means that I need to raise the bar if I want it to replace fdisk,
bsdlabel and sunlabel in the future. Those tools have a UI for
example.

Also, the FreeBSD installer needs to be able to partition disks and
it needs to be able to support all of our schemes. It is itself a
unified tool for partitioning.

So, doing more than the bare minimum is beneficial in various ways.
That's why I'm having difficulty to just replace gpt(8).

-- 
Marcel Moolenaar
xcllnt_at_mac.com