Gerrit P. Haase wrote:
It seems that there is one test failing: TEST params.pas: failed: ./a
Seems like Cygwin strips the `.exe' suffix internally in the 0th argument. (The executable does have this suffix, doesn't it?)
I guess we can just ignore it in the test:
--- test/params.pas.orig Sat Jan 11 15:48:04 2003 +++ test/params.pas Tue Oct 14 00:02:15 2003 @@ -10,7 +10,8 @@ begin s:= ParamStr ( 0 ); if (copy ( s, Length ( s ) - 4, 5 ) = 'a.out') - or (LoCaseStr (copy ( s, Length ( s ) - 4, 5 )) = 'a.exe') then + or (LoCaseStr (copy ( s, Length ( s ) - 4, 5 )) = 'a.exe') + or (s[Length ( s )] = 'a') then writeln ( 'OK' ) else writeln ( 'failed: ', ParamStr ( 0 ) );
And three tests are skipped.
All the ncurses tests are passing well!
So please include the change for cygwin, we can use the ncurses library.
How well does it support special features? See the comment in crt.pas. You can test them with crtdemo (interactively -- there are no automatic tests for them). Perhaps the Chief can report on them with PDCurses.
The compiler/linker output on cygwin seems to be unusual so there are some tests reported as failed which are actual not failing.
I haven't seen this before. Is there perhaps a linker option to suppress these "Infos"?
Frank