Re: [gentoo-user] Re: Gentoo's advantage: 'optimized for your system' -- huh?

[Jesús Guerrero]

El Jue, 5 de Febrero de 2009, 9:11, Nikos Chantziaras escribió:
> Jesús Guerrero wrote:
>
>> El Jue, 5 de Febrero de 2009, 7:07, Nikos Chantziaras escribió:
>>
>>> Sebastián Magrí wrote:
>>>
>>>
>>>> The installation experience with the traditional method must be
>>>> mandatory... That's why I think we are better now that GLI is
>>>> deprecated...
>>> That's not good.  It hurts Gentoo's popularity if it's not easy to
>>> install.  But since there are not enough devs left for the GUI
>>> installer, not much that can be done.
>>>
>>> Gentoo isn't unsuitable for a GUI installer.  It's stage 3, after
>>> all.
>>
>> That's definitely good. It never worked and newbies came to
>> Gentoo thinking that it as some kind of uberfaster ubuntu
>> thing that could be installed by just clicking next.
>
> Than I'll rephrase my statement: Gentoo would need a non-bugged GUI
> installer ;)

I wouldn't have anything against that. But after seeing one
failure after another I think that lots of users are scared
to see yet-another-one that will only make our lives more
difficult.

I wouldn't mind about it if it's developed as experimental stuff
and NEVER ever again included as a valid method of installation
in the handbook unless

A) it's as rock solid as the command line
B) the user ends the procedure knowing the same things
   about gentoo that you would know if you installed by hand
   (i am particularly concerned about this one, and I simply
   can't see how a GUI would accomplish this one at all)

>> Then they ran away yelling how bad this gentoo crap is that
>> doesn't work at all unless you do a lot of black magic on the command
>> line! "Because I want full control over my system, but only clicking
>> next. The OS should read my mind!"
>
> I don't think anyone should care about that.

Well, I only said that because you talked about popularity.
Otherwise, we agree: I don't care at all.

You made some good arguments about GUIs, and I understand them.
We could have a simplified and standardized installer that work
with a standard config. However I don't wanna live yet another
nightmare.

> Also, Gentoo isn't really black magic.  There's no good reason why
> emerge for example isn't GUI based.  Or revdep-rebuild.  Or layman. Or...
> I hope you get the point ;)  Yes, those things need a lot of work
> and there are no people willing to do the task.  But I'm just trying to
> make a point here: the way you do maintenance in Gentoo isn't based on the
> traditional Unix tools.  That means, you could have GUIs for all of them.
>
> But I'm drifting.  The installer is pretty much separated from all this.
> After all, "all" it needs to do is set up stage3 and tweak the settings.

Well. I suppose it's about tastes. But the shell is where emerge
and ebuilds belong for me. After all, the ebuilds are nothing but
bash scripts. You could do frontends to it, but it would still be
a lot of python and bash code behind that. With these tools it
happens the same that with the installer. At one point, tools like
these appear, they are developed for some time and work mostly ok
but not perfect, then they get stagnated, they break more and more
and more with the time, until it comes the day they are unusable
and the project dies.

I guess that -again- because there's zero interest. When you need
to compile something:

A) it can't get any simpler, nicer nor faster than doing emerge
   something, really
B) the last thing you needs is a heavy interface taking
   away your ram and cpu, emerge itself is heavy enough as
   it is, there's no need to add weight to the thing
C) you won't like when X is closed in the middle of emerge
   that's why you run emerges on an vt or a screen session,
   in text mode

And probably many more. I would love, though, to see a curses
frontend where I can dive into my portage dirs in an mc-ish
fashion, which is where portage frontens make any sense for me:
when you just want to take a look around and see what's in there :)

-- 
Jesús Guerrero