Re: Octave 3.6.1 mingw for windows - updated

[Lukas Reichlin]

On 05.04.2012, at 17:09, Jordi Gutiérrez Hermoso wrote:

> On 5 April 2012 10:59, Lukas Reichlin <address@hidden> wrote:
>> 
>> On 05.04.2012, at 15:34, Jordi Gutiérrez Hermoso wrote:
>> 
>>> 2012/4/5 Sébastien Villemot <address@hidden>:
>>>> During the installation it is possible to only install Octave without
>>>> the Octave-Forge packages (by unclicking a checkbox).
>>> 
>>> People are basically illiterate. They don't care what they're
>>> installing or why. They just want "everything". They need a little
>>> guidance here.
>>> 
>>>> From my point of view, if the user installs the Octave-Forge packages,
>>>> he expects them to be autoloaded. At least this is the policy that we
>>>> adopted in the Debian Octave Group.
>>> 
>>> Debian users have a habit to pay attention to what they're installing
>>> and will complain loudly if packages they don't recognise are being
>>> installed. Windows users have the habit of
>>> "next->next->I-agree->next->next->done".
>>> 
>>> - Jordi G. H.
>> 
>> Jordi, I don't quite get it. Why do you want users to type "pkg
>> load" every time if they chose to install the packages?
> 
> No, they can add it to their .octaverc. I think it's important for
> another reason: educate user that some functions are in packages and
> some are in core. And also, to educate users that there is such a
> thing as an .octaverc, in which they can add their customisations.

I don't like the idea of "educating" the users ;-)

> 
> If Windows had a nicer packaging infrastructure, it would be nice if
> you could say "pkg install package" and the package would install
> forever. Sadly, it doesn't, and "pkg install" is essentially useless
> in Windows, so I think "pkg load" (and perhaps patching pkg load to
> add that command to .octaverc?) is the next best thing.

OK, at least "pkg install -forge control" used to work as expected on MinGW.

> 
>> And you don't have to load matlab toolboxes either.
> 
> No, but you have to buy them. It's not unusual for people to write
> Matlab scripts without using certain functions from some toolbox
> because they want to make sure their users can use them regardless of
> which Matlab toolboxes they may have installed.
> 
> I think in this case, trying to make things too convenient for users
> doesn't expose them to the intricacies of the underlying problems, so
> they end up with other problems, such as for example, reporting OF
> problems to the wrong place, or being unaware of the relative quality
> and standards of core Octave and its packages.
> 
> - Jordi G. H.

Aaah, the Octave Core vs. Octave Forge story. I think we should group OF 
packages into different classes as the long list on                             
               http://octave.sourceforge.net/packages.php
is rather confusing. For example, we could use three classes:

- main: only about a dozen of packages. The most popular ones, of general use. 
High quality. Tested and actively developed/maintained. What the average user 
needs, see debian priorities below.

- extra: the less common ones. narrow scope, less tested, under construction, …

- legacy: unmaintained, outdated and fragile ones, like the "symbolic" package.


We had similar discussions about priorities for debian packaging:
        http://www.mail-archive.com/address@hidden/msg06842.html


Then we could use these OF classes for the OF website and the installers. What 
do you think?

Lukas

PS: Sorry for hijacking the thread.