Re: [gtk-list] TODO list: grid widget?




Adrian Feiguin <feiguin@magnet.fsu.edu> writes: 
> This sounds to me very funny, and sad at the same time. Me (and others)
> proposed many times to include GtkSheet in the main distribution,
> specially considering that it is very stable now and include lots of
> interesting features, including bindings to other languages. I even
> didn't recieve any (ANY!) answer from the gtk-team, and this is very
> perturbing, specially after reading the TODO list. What are we doing?
> Aren't we polite enough to communicate? I worked very hard on GtkSheet
> and GtkExtra and I am pretty sure that we can share a lot of experience
> if we want to add such a widget to the distribution, and gain a lot of
> time.
> 
> On the other hand, I have to honestly say that I find very disturbing
> the gtk-team position and attitude, aren't we an open-source community?
> Don't we all participate in the growth of GTK+ as the most amazing, most
> widely used toolkit?
> 

Historically I think you're right that Owen and Tim have not been so
good at getting back to people about patches, but I know that they
have been trying very hard to correct the situation in recent months.

Part of the work of each TODO item is to go out and find existing
code, talk to the people who wrote it, and post information about that
to the list for discussion. GtkSheet might be fine, but the reason
a sheet widget is on the TODO list is that no one has gone out and
done the work to evaluate GtkSheet and look at getting it in GTK+.

In general one might post to gtk-devel-list with:

  - design goals for the widget; in particular, this should include
    _clear limits_ on the functionality the widget will
    have. Featuritis is very bad for core GTK widgets.

  - the API you have written or will write

Then, you need to iteratively work to get the widget in GTK, for
example:

  - respond to any concerns that arise on the mailing lists

  - make changes the GTK maintainers request
 
  - make sure the whitespace and comments please Tim the format
    fascist ;-)

  - mail the list and/or Owen and Tim and bug them if they 
    owe you a response and haven't responded

Anyway, that's the general stuff. 

Some specific comments on GtkSheet, based on looking at it
quickly. These are things I figure would come up if it was discussed
here:

  - it isn't model/view
  - freeze/thaw are kind of a bad API feature, if possible it's much
    better to simply make all of the operations fast
  - the SheetButton/SheetCell/etc. structs shouldn't
    be exposed in the header file
  - it should probably have an infinite scroll area, 
    which is free in GTK 1.4, but requires GtkLayout-style
    tricks with 1.2
  - formatting issues: functions named in all caps, 
    braces in the wrong place, etc.
  - object arguments aren't implemented
  - key navigation should use GtkBindingSet

Questions that maybe need to be addressed:
  - how does GtkSheet compare to the Qt sheet and the Java sheet?
    How does your design differ and why?
  - what are the performance and memory properties of GtkSheet?
    (that is, can you efficiently have a large sparse sheet, 
    how slow/fast is it to insert items in the middle, this 
    sort of thing)
  - what are the intended applications of GtkSheet, and does 
    the current feature set match those applications? what 
    features have explicitly been _rejected_ from the widget, 
    as too complicated or too bloated or too useless?

If you have time to do this extra work on GtkSheet, it would be a
great contribution to GTK, IMHO.

Havoc




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