Re: Some remarks to pgadmin III - Mailing list pgadmin-hackers

Adam H. Pendleton wrote:

> Andreas Pflug wrote:
>
>> Adam,
>> this is a waste of time. DATA IS NOT COMPLETELY ON THE CLIENT, as I
>> stated earlier. The grid doesn't have a chance to sort unless all
>> data is loaded, but it's just the special feature that data is *not*
>> loaded completely!
>
>
> What data exactly isn't at the client?

All data that's never been displayed.

>   If the sorting of the data that *is* at the client (the values
> displayed in the grid are most certainly rows inside the grid control
> inside the application running on the user's PC) is implemented at the
> wxWindows level and is completely transparent to pgadmin3, what
> difference could it possible make?

Adam, I believe it's time you have a look at the source what's really
going on there. The grid doesn't have control over the data, it's
pgadmin3 managing it.

>
>
>> Other libraries don't implement this either. You'd have another
>> problem: Which key code is going to be "Select all"? Is it the one
>> that seems to be used more or less widely on the platform,
>> Apple-A/Ctrl-A/Alt-A, or the one that many editors use, including the
>> embedded Scintilla which implements a complete set of keyboard
>> shortcuts? I doubt that the wx people would accept any of both
>> solutions.
>
>
> As I said in my previous e-mail, the key sequence and the behavior
> isn't driven by the library, it's driven by the control/widgets that
> wxWindows uses to implement its dialogs in a native format.  For
> example on Linux, the select all text would work if the GTK+ widget
> for a text box supported it.  The same with the comctl32.dll controls
> on Windows.  This is the whole point of a window manager implementing
> widgets for its application to use: common behavior across all its
> applications without having to write code to handle it at the
> application level.   Now if Scintilla wants to override this default
> window manager behavior, and implement its own shortcuts, then more
> power to it, but it doesn't change anything with regard to what we're
> talking about.  All I'm saying is that the standard window manager
> behavior should be supported and implemented, if possible (and it
> shouldn't require any code by us).
>
> Take, for example, the Minimize/Maximize/Close buttons in the
> upper-right corner of every Windows application.  You don't implement
> the code to create these boxes, specify their behavior, handle their
> events, etc.  This is done by the window manager.  You just create  a
> window, tell the window manager whether you want those boxes to
> appear, and that's it (well the Close box event gets handled, but its
> a special case).  You also don't implement code to handle movement of
> the window around the screen, except for the regular re-size, re-draw
> events, and then only if you have special circumstances.  This is the
> exact same thing as supporting Ctrl-A,Alt-A, or whatever the standard
> "Select All" key combination is for the particular window manager.
> NOTE: I am definitely saying that implementing this should*NOT* be
> done at the pgAdmin3 level, unless absolutely necessary, and if it
> does become necessary then we can argue over which key combinations to
> use.
>
Minimize/Maximize is implemented by GDI and gtk, Select All is not. You
want to change the Win API? Good luck....
BTW, Alt-A is quite a bad idea, Alt keys should be left free for
application shortcuts, not used by controls.

Regards,
Andreas


> ahp
>
>>
>>
>> Regards,
>> Andreas
>>
>>
>>
>
>
>