Hi,

here is new qt patch.

+ Rewritten TQtWSCustomFloatSpinEdit, so we have now fully useable TSpinEdit and TFloatSpinEdit
+ Added OnChange events to TCustomEdit and TCustomMemo
+ Added RoundRect() to winapi.

cheers,
zeljko
-------------- next part --------------
A non-text attachment was scrubbed...
Name: qtupdate-r11237.patch
Type: text/x-diff
Size: 16183 bytes
Desc: not available
URL: <http://lists.lazarus.freepascal.org/pipermail/qt/attachments/20070531/03420580/attachment.bin>

Thanks, applied

Hi zeljko,

Are you porting a large software to the qt interface, or something like that? Just curiosity =)

I keep wondering how far is the qt widgetset from being able to run lazarus IDE. This would be definetively great =)

thanks,

Indeed.

What about 'native' qt on MacOSX?
Are there any differences between lcl qt for the various platforms?

Mattias

Felipe Monteiro de Carvalho wrote

Hi zeljko, Are you porting a large software to the qt interface, or something like that? Just curiosity =)

Nope, I have some free time those days (till Sunday), so I'm trying to help as much as I can :), and of course maybe I'll port my software to lazarus-fpc one day, but not now (MDI support missing).

I keep wondering how far is the qt widgetset from being able to run lazarus IDE. This would be definetively great =)

IDE can run, but there's some error with adding menu actions, so it dies at that point (screenshoot attached - I have to scale image, only 40k msg allowed). I think that it's not so far away from
coding under qt lazarus.

* Is there any chance to commit latest qt4.pas from Den (with QStyle) added so we can use QStyle drawing utilites (qt-4.2.3) ?

* I've installed FPC-2.1.4-beta, and same problem is still sitting at qt4.CopyUnicodeToPWideString() just raises AV when any text change occurs in QLineEdit and QTextEdit (those events are implemented with my latest patches). AV raises when Len parameter is > 1 ?!? I think that this is FPC bug.

As I said workaround is to set try..except around Setlength(s, Len) inside this routine, and then everything looks fine:

try
SetLength(s, Len);
except
  exit;
end;

cheers
zeljko

-------------- next part --------------
A non-text attachment was scrubbed...
Name: qtlazarus.jpg
Type: image/jpeg
Size: 21961 bytes
Desc: not available
URL: <http://lists.lazarus.freepascal.org/pipermail/qt/attachments/20070531/08029cd3/attachment.jpg>

Mattias Gaertner wrote

What about 'native' qt on MacOSX? Are there any differences between lcl qt for the various platforms?

Qt is by far the library with the smallest amount of cross-platforms problems from all that I tested so far. So, if something is running with lcl-linux-qt, I would give it 99% certainty that it runs equaly good with lcl-mac-qt

We have noticed some small problems on qt-mac on the past, but they were minimal as far as I remember, so nothing to worry about.

zeljko wrote

IDE can run, but there's some error with adding menu actions, so it dies at that point (screenshoot attached - I have to scale image, only 40k msg allowed). I think that it's not so far away from coding under qt lazarus.

Is the code editor that we see on the background running with qt? That would be quite impressive ...

yes, it's TSynEdit, I think that we need to implement needed winapi painting functions and it'll work :), but we need also QStyle in qt4 bindings ... (newest qt4.pas from Den) ;)

* Is there any chance to commit latest qt4.pas from Den (with QStyle) added so we can use QStyle drawing utilites (qt-4.2.3) ?

I updated our qt bindings to version 1.30, please test.

As I said workaround is to set try..except around Setlength(s, Len) inside this routine, and then everything looks fine:

I commited the workaround, but, considering this problem, it may be better to simply not use this function.

I would be better if fpc provided functions to work with widestrings

Than we cannot continue our work on qt4 widgetset. Until better solution this one is harmless (I think so). :), and we cannot bypass this function since qt4 bindings use it (or yes maybe ..?!? if we use directly QStringH (but not yet inside qt4)).

I would be better if fpc provided functions to work with widestrings

fpc have functions like SetString() etc... but raises segfault also... don''t know why , but always when trying to setLength() > 1 , while SetLength(s, 1) works, and only crashes with qt4 events (textChanged).

cheers,
zeljko

zeljko wrote

Than we cannot continue our work on qt4 widgetset. Until better solution this one is harmless (I think so). :), and we cannot bypass this function since qt4 bindings use it (or yes maybe ..?!? if we use directly QStringH (but not yet inside qt4)).

oh .... in this case I would say our workaround is ok =)

fpc have functions like SetString() etc... but raises segfault also... don''t know why , but always when trying to setLength() > 1 , while SetLength(s, 1) works, and only crashes with qt4 events (textChanged).

Please submit a bug report to FPC's bugtracker, and tell them you tested with the beta too. A small test program would be great too. It significantly diminishes the time to fix bugs.

thanks,

Felipe Monteiro de Carvalho wrote

Please submit a bug report to FPC's bugtracker, and tell them you tested with the beta too. A small test program would be great too. It significantly diminishes the time to fix bugs.

hm...it's hard to reproduce since SetLength() works ok in normal conditions, think that fpc guys should test it directly with qt interface while textChanged_events being called ... impossible to put it into small test program :)

Just a wild guess:
Maybe somewhere a PWideString is typecasted to WideString.

Mattias

that was first thing I was searching for, but didn't found anything bad.