Hi,
Sven Neumann wrote:
> Dave Neary wrote:
> >> And page 5 (monitor calibration) should definitely be skipped. Get
> >> the resolution my default from the monitor, as we do, and allow
> >> manual calibration in the preferences.
>
> The question here is "Why?". I don't see any good reason to skip this
> page. The monitor resolution needs to be set, why do want to force the
> user to find out where she can do this? Doing it during the user
> installation is a perfectly good place to present the user with this
> choice. I do not at all understand the reasoning behind simpliying the
> user installation. GIMP is not a simple app, we can not and we do not
> want to hide it's complexity. Why should we?
The assumption here, and elsewhere, is that most users will know
what is being asked of them, and that many of them will care, and
want to modify the default settings proposed.
I guarantee you that that is not the case. Most GIMP users click
OK, Yes, OK, OK, OK, with a total time elapsed of less than a
minute. Most GIMP users don't even read the tile cache dialog, or
the screen calibration dialog.
The intro screens (with the exception if the tile cache,
arguably) propose decent defaults. The whole point of the intro
screen is (should be) to allow the user to modify things global
to the installation (the directory in which the app installs
globally on Win32, for example, or whether feature X should be
enabled).
Proposing modificatiopn of user preferences in an install procedure
is generally something which just gets in the way. Particularly
when the defaults are grand.
As for the particular example of the screen calibration, it does
need to be set. But setting it to the default we get from X is
satisfactory in 99% of usage situations. And if it's not
satisfactory, we can assume a fairly high level of knowledge for
the user (if he knows that his screen calibration is off, he's
probably not the average linux user), in which case allowing
manual changes in the preferences is not hiding it away.
To answer the more general point (the GIMP is complicated), I
would argue that the complexity of the application is independent
of the knowledge level of the user, and of the complexity
required for installation. We have a complicated preferences
dialog which reflects a lot of user-configurable behaviour, yet
we don't ask the user to choose which interpolation method he
will use when scaling an image during the installation. We set a
decent default, and if he knows enough to notice a difference,
there is the possibility to change it. This is, IMHO, a
reasonable approach.
I don't really get the argument that the users home directory
is sacred, and we shall not put files in there without asking
for permission first. My wife is not a power user, but she
occasionally uses quite a few apps. In her home directory ls -ad
.* | wc -l gives 87 dotfiles (including . and ..) - many of those
are directories which install a local profile (kde, gphoto,
mozilla, galeon, openoffice, etc). Many of those are complex
applications. Most of them do not inform the user that they're
creating files in the home directory, or if they do, they do so
with a dialog that says "Creating user profile... Done", and
nothing else.
That's not to say that just because everyone else does that, so
should we. Elsewhere, you have said "Unlike GNOME, we don't
assume the user is stupid". I'd argue that a graphics app should
assume that the user is stupid (by which I mean, unfamiliar with
the inner workings of the computer). I would argue that setting a
reasonable tile cache, getting the screen calibration from the
windowing system, and putting the swap file in either .gimp-1.3
or /tmp (why don't we put it in /tmp?) is fine for the vast
majority of GIMP users. If that's the case, there is no need to
ask permission, and no need for them to know what you are doing
in detail.
What's being proposed is replacing the GIMP installation
procedure by something which conforms more to the average users
expectations - personally I think that's a reasonable goal.
Cheers,
Dave.
--
David Neary,
Lyon, France
E-Mail: bolsh@xxxxxxxx
