[Lazarus] Does using GtkFrame in LCL make sense?

[Max Vlasov]

Hi,

I recently created report http://bugs.freepascal.org/view.php?id=19658

This was about panels that behave a little differently to panels in Lazarus
Windows and Delphi. There were no comments or feedback request there so I
decided to take a look

After some investigation it appeared that the problem was because GtkFrame
was used. The problem with this widget is that one can't control the postion
and painting in a small area two pixels around it. So every nested TPanel in
GTK2 Lazarus smaller by 4 pixels that the parent no matter what (the
consequences can be seen in the screenshots from Panel.zip). Looking at the
code I also noticed that no special features of GtkFrame were used, all
panel drawings take place inside paltform-independent part of LCL.

Trying to solve the problem, I ended up with the following replacement in
the TGtk2WSCustomPanel.CreateHandle

{ -------- Before }

  Frame := gtk_frame_new(nil);
  gtk_frame_set_shadow_type(PGtkFrame(Frame),
    BorderStyleShadowMap[TCustomControl(AWinControl).BorderStyle]);

{ -------- After }

  Frame:=gtk_scrolled_window_new(nil, nil);
  gtk_scrolled_window_set_shadow_type(PGtkScrolledWindow(Frame),
    BorderStyleShadowMap[TCustomControl(AWinControl).BorderStyle]);
  gtk_scrolled_window_set_policy(PGtkScrolledWindow(Frame),
GTK_POLICY_NEVER, GTK_POLICY_NEVER);

After this change the problem project that looked identically (good) in
Delphi and Win32 Lazarus and differently in GTK2 Lazarus (labels and panels
at wrong positions) began to look consistent in linux.

So there are questions:
- Can this patch be applied? Apart from obvious testing and checking, there
is also a compatibility problem. After this change, the forms people
designed in linux can start to look a little different (or even greatly if
the nesting level of panels is deep). But as I noted in the report, without
this change it's difficult to synchronize forms "visually".
- Does using GtkFrame in LCL make sense at all? I did not look deeper, but
at least quick search showed it was used several times.

Thanks,

Max Vlasov
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus-ide.org/pipermail/lazarus/attachments/20110701/c5f42081/attachment-0002.html>