Hello,

in new SVN revision (31941) from this morning, compilig Lazarus appears this error (In Win32):

Compiling syneditlines.pas
syneditlines.pas(57,5) Note: Private field "TSynEditLines.FLineEndType" is never used
synedit.pp(4710,12) Error: Wrong number of parameters specified for call to "WMSetFocus"
synedit.pp(9000) Fatal: There were 1 errors compiling module, stopping Fatal: Compilation aborted

Regards,
Juan.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20110811/7f1d3a2c/attachment.html>

2011/8/11 Juan S?nchez :

What fpc version do you use? I am aware of some issues using fpc 2.7.1, because of
http://wiki.freepascal.org/User_Changes_Trunk#Anonymous_inherited_calls

Vincent

2011/8/11 Vincent Snijders

Maybe because FPC. I'm using FPC 2.7.1 from SVN updated today. Yerstarday it compiled OK with 2.7.1 but today not.

Juan.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20110811/21f5095e/attachment.html>

Indeed, I found that with FPC 2.4.4 the error does not occur.

The problem seems to be FPC 2.7.1
Juan.

2011/8/11 Vincent Snijders

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20110811/96cb227f/attachment.html>

Juan S?nchez wrote

Indeed, I found that with FPC 2.4.4 the error does not occur.

Bottom line... I don't think using unstable (trunk) versions from two products together is always such a wise choice. ;-)

Regards,
- Graeme -

What fpc version do you use? I am aware of some issues using fpc 2.7.1, because of http://wiki.freepascal.org/User_Changes_Trunk#Anonymous_inherited_calls

Can you explain what this really means?

I don't understand "anonymous", "call through", and the example :-(

IMO a compiler should reject ambigous overloaded procedures, i.e. when the significant parameters only differ in type size.

I understand that there was a problem, when files larger than 4GB occured in Windows, so that TStream.Size and .Position had to be updated from LongInt to Int64. The choice of LongInt already was a questionable decision, because it would restrict the file size to 2GB. When the Delphi solution requires overloaded methods, i.e. in TStream:

     procedure SetSize(NewSize: Longint); virtual;overload;
     procedure SetSize(const NewSize: Int64); virtual;overload;

we are in trouble, since it's up to the compiler to choose one of these methods - based on what criteria? :-(

DoDi

anoymous -> "inherited" instead of "inherited SomeCall(AParam: Type);"

call through -> well, I'm no native English speaker either, and I also found "through" isn't an appropriate word. IMVHO it's the new behavior that calls "through" the parent into the exact overriden method ;-) The example is clear for me. Anonymous 'inherited' will call an exact inherited implementation instead of a compatible overload in the parent/intermediate ancestors. I wasn't really aware of the old/current-release FPC behavior :-o

IMO a compiler should reject ambigous overloaded procedures, i.e. when the significant parameters only differ in type size.

It doesn't reject the overloads?

Best regards,
Fl?vio

Vincent, on a side note, are you sure this applies to 2.7 and not 2.5/2.6? I believed the IUnknown changes had been in trunk for a long time and would be in 2.6, won't they?

Best regards,
Fl?vio

2011/8/11 Fl?vio Etrusco :

Vincent, on a side note, are you sure this applies to 2.7 and not 2.5/2.6? I believed the IUnknown changes had been in trunk for a long time and would be in 2.6, won't they?

What does IUnkown has to do with it?
Besides, the snapshots at
ftp://ftp.hu.freepascal.org/pub/lazarus/snapshots/ seem to indicate this is a recent issue.

Vincent

Those changes of User_Changes_Trunk that are already part of 2.6 will be moved to User_Changes_2.6.0 (or so) when someone of the FPC devs finds the time to check/move them.

Regards,
Sven

> I don't understand "anonymous", "call through", and the example :-( anoymous -> "inherited" instead of "inherited SomeCall(AParam: Type);"

Okay, that makes sense.

call through -> well, I'm no native English speaker either, and I also found "through" isn't an appropriate word. IMVHO it's the new behavior that calls "through" the parent into the exact overriden method ;-)

Makes sense, too.

How else would the example compile?

test(b: byte) is not really distinguishable from test(b: longint).

DoDi