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

Re: "task" and beyond, aptitude is not solution

#include <hallo.h>
* Osamu Aoki [Fri, Feb 07 2003, 09:13:51PM]:

> As for 3, finding ssmtp or nullmailer is not easy.  Also postfix should
> given position of alternative too.
> Current package threshold of 
>  Req
>  Imp
>  Std
>  Opt
>  Ext
> is too rough.  Also these nor section can not be used as screening
> methods.

Insufficient nowadays.

Req: 41 (0.4%)
Imp: 41 (0.4%)
Std: 81 (0.7%)
Opt: 9090 (82.1%)
Ext: 1819 (16.4%)

> Most programs are Opt.  Also some useful programs (ssmtp, nullmailer)
> are (or should be) in hidden in Ext by policy due to its conflicts.
> Unless we have nice alternative selection method, it is unpractical to
> expect aptitude can solve problem.

Exactly. Most maintainers throw everything into optional because few
packages in Std "define" what the "core" is and everything with less
importance (for administrator) has to be below. Why do a "desktop" user
need bind9-host in the first line? Or bison? Or cpp? We should realise
that if Debian becomes more user oriented, the users will need to
priorize a range of packages and be able to set the threshold so all
packages below it are hidden.  The (pre)selection of the package series
could be done by Subprojects (-demudi, -desktop, -jr). What I suggest is
similar to the current task scheme and could replace the task-scheme,
imo. When the users set the focus on certain package series and says:
install all of _my_ Standard packages, she will get it. Otherwise she
can browse through the minimised package selection and choose what she
And yes, this will _definitely_ require us to set more detailed
preference levels on different packages. Some maintainers would cry:
"Why is my package less important than nedit? No,no,no." but its time
accept the truth: the user does not care about 30 editors and 20 http
servers, he wants to find _her_ _one_ software packages, and she wants
to get it easily, and it should _just work_. 
That is one of the reasons why M$'s wanna-be-OSs have so much success.
You install it, and you get an editor, a calculator, etc.  It is here.
It is easy to start with it. You do not need 15mins to find the package
you need (say nedit, gcalc). If you need a more advanced editor, you can
browse around and find gvim, xemacks, etc. pp. But only then and not
from the beginning. 
And the "support packages" should get the lowest priority then.

Everything should be made as simple as possible, but not simpler.
                -- Albert Einstein

Reply to: