On Mon Apr 09 23:01:35 2007, [EMAIL PROTECTED] wrote: > On Sunday 08 April 2007 18:07, Steve Peters via RT wrote: > > > On Sun Apr 08 16:08:05 2007, stmpeters wrote: > > > The attached patch includes several cleanups needed to silence > > > warnings > > > when compiling Parrot with Intel C++. > > > > It helps to attach the right patch.... > > I get several warnings. I've cleaned up this batch: > > src/pmc/eval.pmc: In function ‘Parrot_Eval_get_string’: > src/pmc/eval.pmc:255: warning: passing argument 3 of ‘PackFile_pack’ > from > incompatible pointer type > src/pmc/eval.pmc: In function ‘Parrot_Eval_thaw’: > src/pmc/eval.pmc:312: warning: passing argument 3 of ‘PackFile_unpack’ > from > incompatible pointer type > > src/pmc_freeze.c: In function ‘run_thaw’: > src/pmc_freeze.c:1435: warning: comparison of distinct pointer types > lacks a > cast > > src/pmc/string.pmc: In function ‘Parrot_String_nci_trans’: > src/pmc/string.pmc:853: warning: array subscript has type ‘char’ > > ... but my attempt to fix these causes more test failures in > t/op/string_cs.t: > > src/encodings/fixed_8.c > src/encodings/fixed_8.c: In function ‘get_byte’: > src/encodings/fixed_8.c:49: warning: pointer targets in initialization > differ > in > signedness > src/encodings/fixed_8.c: In function ‘set_byte’: > src/encodings/fixed_8.c:67: warning: pointer targets in assignment > differ in > sig > nedness > src/encodings/ucs2.c > src/encodings/utf16.c > src/encodings/utf16.c: In function ‘get_byte’: > src/encodings/utf16.c:170: warning: pointer targets in initialization > differ > in > signedness > src/encodings/utf16.c: In function ‘set_byte’: > src/encodings/utf16.c:188: warning: pointer targets in assignment > differ in > sign > edness > src/encodings/utf8.c > src/encodings/utf8.c: In function ‘to_encoding’: > src/encodings/utf8.c:334: warning: pointer targets in assignment > differ in > signe > dness > src/encodings/utf8.c:357: warning: pointer targets in assignment > differ in > signe > dness > src/encodings/utf8.c: In function ‘get_byte’: > src/encodings/utf8.c:400: warning: pointer targets in initialization > differ > in s > ignedness > src/encodings/utf8.c: In function ‘set_byte’: > src/encodings/utf8.c:418: warning: pointer targets in assignment > differ in > signe > dness > > The test results are: > > not ok 5 - downcase > # Failed test (t/op/string_cs.t at line 72) > # got: 'aeiou_�� > # ' > # expected: 'aeiou_�� > # ' > ok 6 - upcase > not ok 7 - titlecase > # Failed test (t/op/string_cs.t at line 90) > # got: 'Zaeiou_�� > # ' > # expected: 'Zaeiou_�� > # ' > > As seen through less, they are: > # Failed test (t/op/string_cs.t at line 72) > # got: 'aeiou_<C4><D6><DC> > # ' > # expected: 'aeiou_<E4><F6><FC> > # ' > # Failed test (t/op/string_cs.t at line 90) > # got: 'Zaeiou_<C4><D6><DC> > # ' > # expected: 'Zaeiou_<E4><F6><FC> > > ... so the encoded characters get 32 added to them somehow, somewhere. > > I've attached your patch with a few changes on my end. > > -- c > >
These are likely caused by the varying definition of what a STRING->strstart is. Sometimes its a char *. Sometimes its an unsigned char *. The pointer itself is a void *. Its a big mess. Obviously, this all needs more work, and probably a bit more thought on my part. I'll probably break apart this patch to get the enum fixes in and deal with the larger STRING issue separately. Steve
