ANN: KGrid 1.4

[DRIGUS]

Hi TK.

I already stepped back to Canvas.Textout but I'll make a conditional directive now, because I think the normal Textout is still little bit faster.

  In Windows 32 TextOut is a Windows API function call, but TextOut is missing in Windows CE. Therefore I used ExtTextOut without calculating the bounding rectangle. In order to achieve the same results as TextOut it must be modified (see bugtracker issue for more information). I have no idea if that has any consequence on painting speed.

And I'll keep the alphablended arrows unchanged (unless somebody reports a problem on Carbon).

  Until the fix is integrated another workaround is to modify TKAlphaBitmap.AlphaDrawTo and use your own blending function (combine the defines FPC and not WINCE).

One question yet about the performance on WinCE: Does it paint slow even without the inplace editor (goEditing inactive)?

  The problem is not the painting speed, for example there is no noticeable flickering. It is more related to the reaction time: When you click another cell it takes approximately 500 ms until you get a visible feedback. In combination with the inplace editor you get the following behavior after selecting a new cell: It takes around 500 ms until the inplace editor is visible in the new cell and after that a noticeable short time until the old editor disappears.

Greets,
  Björn

[tk]

Hi,
ok then I'll consider to use my blending function for WinCE too - I remember it was not working for GTK2 and maybe QT too even if I chose another pixel format for TLazIntfImage (32bpp without alpha channel).

If I understand you well painting speed and reaction time without inplace editor (goEditing=False) is OK but the reaction time e.g. while changing the inplace editor to a new cell is slow. Yes this is exactly what I can experience on GTK2 now (yep it is faster here than those 500ms but it is slow though). Well, I don't know how to speed this up for these widget sets. It might be a problem because I set several TWinControl properties (I have to set these), I get RTTI info for some etc. Perhaps some update lock would help in the InternalEditorxx functions but no such lock exists in the LCL afaik.

TK

[DRIGUS]

Hi TK.

If I understand you well painting speed and reaction time without inplace editor (goEditing=False) is OK but the reaction time e.g. while changing the inplace editor to a new cell is slow.

  No, there is no noticeable difference in painting speed regarding goEditing. The only difference is that you can see how the old inplace editor disappears after the new one got visible when using goEditing.


Greets,
  Björn

[tk]

Hi all,
I've released new version 1.5 with robust printing and previewing capabilities. It is still beta release as the printing works only for Win32 because of the missing affine transformations in other widgetsets (I'll try to implement these for some other Lazarus widgetsets later).

To Björn about WinCE:
Now is Canvas.TextOut by default for Lazarus but can be switched off using a kcontrols.inc directive. I removed the functions with Extended in Lazarus. I didn't modify the alpha glyphs under WinCE - they rely on TLazIntfImage in other widget sets than Win32. TLazIntfImage must work here as expected, i hope your patch will be applied soon. Painting is yet "little bit" faster, as I see in Linux/Gtk2 now. I think the issue with inplace editor cannot be solved unless LCL is more optimized for this. I made a benchmark program Tstringgrid/Tkgrid in GTK2 and with the same amount of painted data kgrid is even little bit faster. If you want to use fast range selection specify the rsMS_Excel option.

Regards,
TK

[CC]

Hi TK,

I have been looking for a cross-platform and "cross-IDE" grid for a long time. KGrid is the first really promising candidate. I only hope you can solve the WinCE issues soon enough for my current project.

Good Luck,
CC
  

[tk]

Hi,
I still have no device to test it but currently I am not aware about a major problem on WinCE. The only is the inplace editor slow reaction (the same can be observed with stringgrid and afaik cannot be fixed) and the fixes made by Björn to the LCL (must be applied by Lazarus team). If you experience some other problems please report (better on my web).

Regards,
TK.

[CC]

Thanks!

[avra]

I wanted to download but site was down and Torry has bad link, so after a little searching here it is: http://www.filesmeethere.com/File/ak0dcm9io09pg. Maybe someone will find it useful.