Re: [PATCH] gnustep-gui cleanups

[Serg Stoyan]

Hello Nicola,

NP> A small comment - the following warnings are spurious, and are
NP> generated by a bug in the compiler.  Casts from (id <GSPasteBoard>)
NP> to (id) were broken in GCC compilers before 3.3, which would ignore
NP> them.  I fixed these casts on GCC CVS HEAD on Sep 17 2002.  So, to
NP> remove those warnings we don't need to change the protocols ... but
NP> upgrade the compiler :-)

[snip]

Maybe add this somewhere in documentation? Something like: "If you see
warnings during compile process and bother this, try to upgrade the
compiler to 3.3 or later". ;) However no gcc 3.3 released yet. :(

NP> About the remaining warnings,
NP> 
NP> NSOpenGLContext.m: In function `+[NSOpenGLContext currentContext]':
NP> NSOpenGLContext.m:67: warning: multiple declarations for method
NP> `currentContext'../Headers/AppKit/NSOpenGL.h:95: warning: using
NP> `+(NSOpenGLContext *)currentCont ext'
NP> ../Headers/AppKit/NSGraphicsContext.h:163: warning: also found
NP> `+(NSGraphicsCont ext *)currentContext'
NP> 
NP> I think this is an error in the API.  -currentContext should return
NP> 'id', for the same reason that -init returns 'id'.

In the attached patch this is fixed. NS*Context entries are changed to
id. I see no harm in such changes.

NP> NSPrintOperation.m: In function `-[NSPrintOperation
NP> _printOperationDidRun:succes s:contextInfo:]':
NP> NSPrintOperation.m:585: warning: passing arg 2 of
NP> `performSelector:withObject:wi thObject:' makes pointer from integer
NP> without a cast
NP> 
NP> While this is an error in the code.  'success' is a BOOL, and can't
NP> be passed as argument as if it were an object.ess]

I've changed [delegate performSelector:withObject:withObject:] to
calling of didRunSelector IMP. It seems to be correct but needs
checking.

-- 
Serg Stoyan

gui2.patch
Description: Binary data