Re: [Gimp-developer] Interaction design. Used to be: Re: gimp gradients

[Michael Natterer <mitch gimp org>]

Hi Peter,

I think it's best to not reply the the thread, because I have troubles
following it myself already, and I better describe how I have
experienced the UTT situation, and the unfortunate conversation with you
on IRC several weeks back.

During last year's GSoC, Mikachu has repeatedly approached me with
problems about how to implement the specified GUI, and it was clear to
both of us that there is some under-specification going on and that we
simply didn't have had enough spec iterations to make it fully
implementable.

There was no hurry at the time, and I had really little time to attend
to it, so I told him to postpone that part and instead get the internal
appraratus really sound, and deal with the UI when we (Mikachu, Peter,
Mitch) have had the chance to discuss the issues and get the spec
updated for corner cases where is simply was not implementable as-is.

The unimplementable part was that the spec didn't say what's happening
for angles where your drawing wouldn't work, and it was clear that the
drawing code would explode in our faces, so I decided to tell Mikachu
not to do any evil hacks that couldn't be right, and leave the GUI
simple until you came back.

Also, there is the issue with the dialog with an infinite amount of
buttons. That's simply because it was somewhat unclear how to integrate
that into general image undo, and kill the dialog for good, which will
clearly happen before 2.10, but we didn't see it as immediate problem
because the undo/redo stuff was there, and the dialog can just be moved
away as long as it exists in 2.9, and it won't be an issue in 2.10
anyway. In fact, one can simply pretend the dialog was not there and not
worry about it.

When you came to IRC in Feb/March I think it was, I was really excited
to finally discuss this and some other open issues with you, and we had
a really nice talk for like an hour or so, until I brought up the UTT. I
don't know what exactly triggered the miscommunication, but what I meant
to say was this:

- The interaction is *completely* implemented as specified
- So are the handles (not visually)
- So are the constraints
- So is undo/redo
- We even took care of handle-specific mouse cursors that rotate
  along with the transform frame

What is NOT implemented:

- The handles are completely not as specified, because we had no clue
  how to implement them for extreme angles, and we were hoping for
  your input
- The dialog is still there, which is temporary

And I also had a change request for the spec:

- The tool is so insanely useful, also for corrective transform and
  really precise working, that it would be a shame to remove the grid
  options. In fact, it's IMO the only usable tool now, so minus number
  entry for rotate/scale, there is actually not any need to have other
  tools.

Which I meant as a compliment, because the spec, the interaction, and
the constraints are totally the shit. I even used the words "I love it",
that's hardly disrespect in my opinion.

Now for general issues:

I have thought hard about last years meeting in Vienna, and all I
remember was trying to mitigate a confrontation between you and pippin,
and not engaging in confrontation. You can't expect me to completely
side with you when in my opinion none of you guys was completely right
or wrong.

Also, about GIMP sliding back from a project where good UI matters to an
engineering project where we don't give a shit about interaction
design... You know, that really hurts. How did that perception happen?

It actually happens *all* the time that we hit a problem with the GUI,
and when you are not around I miss you every time, because what are we
supposed to do? Not make a change that is technically needed (remember
we are in the process of completely changing the inner core of GIMP to
GEGL)? Of course not. Instead, we do it as good as we can, and hope for
your input when you are around next time.

You might also have noticed that we're trying hard to fix issues in the
specs that were implemented in the past, in particular Save/Export and
single-window-mode. We take bugs serious and try to get it right,
sometimes to our best knowledge in case the spec missed some corner
case. And we defend the specs we have all agreed about to an insane
amount of bitching, doesn't that prove our respect for your work?

Also, what am I supposed to do when somebody comes with a new feature,
but we have no spec for its GUI yet? I can't tell everybody to wait
until there is a spec (and we do that often enough). Instead, we do our
best to make it as good as possible anyway. We can't have each and every
GUI change block on the actions of one single entity.

tl;dr

So please, can't we just put all that weird disagreement and
misunderstanding behind us and work together again? I try the same,
because I can tell you, I am not exactly amused about the current
situation either, in particular not about the outcome of our last
discussion on IRC. That has let me standing in the rain, despite my best
efforts to explain the UTT situation. The abrupt end of that
conversation was not what I had hoped for, and I'm most certain the same
is true for you.

You know how much I hate writing mail, and I'm on the 3rd page already
in my editor, that should be evidence enough that I mean it ;)

Regards,
--Mitch