Problems with Revision 6013 (MINIX)

[pikpik]

Hi,

I had a "bleeding_edge" revision from several weeks ago building
successfully on MINIX. However, with revision 6013, I've had some
problems.

Problems include:

 - va_list is missing when compiling some files.
The solution I used involved adding "#include <stdarg.h>" to the
complaining file.

- PrintF seems to want an argument be an "unsigned int" while several
parts of V8 are passing variations of "int's."

scons: Reading SConscript files ...
scons: done reading SConscript files.
scons: Building targets ...
/bin/g++ -o obj/release/deoptimizer.o -c -Wall -Werror -W -Wno-unused-
parameter -Wnon-virtual-dtor -pedantic -I/usr/pkg/include -m32 -O3 -
fomit-frame-pointer -Wno-char-subscripts -fno-strict-aliasing -I/usr/
pkg/include -D_POSIX_SOURCE -fno-rtti -fno-exceptions -I/usr/pkg/
include -Wall -Werror -W -Wno-unused-parameter -Wnon-virtual-dtor -
pedantic -I/usr/pkg/include -m32 -O3 -fomit-frame-pointer -Wno-char-
subscripts -fno-strict-aliasing -I/usr/pkg/include -D_POSIX_SOURCE -
DV8_TARGET_ARCH_IA32 -DENABLE_VMSTATE_TRACKING -
DENABLE_LOGGING_AND_PROFILING -DENABLE_DEBUGGER_SUPPORT -Isrc src/
deoptimizer.cc
cc1plus: warnings being treated as errors
src/deoptimizer.cc: In member function 'bool
v8::internal::Deoptimizer::DoOsrTranslateCommand(v8::internal::TranslationIterator*,
int*)':
src/deoptimizer.cc:780: error: format '%08x' expects type 'unsigned
int', but argument 3 has type 'int32_t'
src/deoptimizer.cc:785: error: format '%08x' expects type 'unsigned
int', but argument 3 has type 'int32_t'
scons: *** [obj/release/deoptimizer.o] Error 1
scons: building terminated because of errors.


Thanks,
pikpik

[Stephan Beal]

On Tue, Dec 14, 2010 at 10:42 PM, pikpik <pikpi...@gmail.com> wrote:

- PrintF seems to want an argument be an "unsigned int" while several
parts of V8 are passing variations of "int's."

...src/deoptimizer.cc:780: error: format '%08x' expects type 'unsigned

int', but argument 3 has type 'int32_t'
src/deoptimizer.cc:785: error: format '%08x' expects type 'unsigned
int', but argument 3 has type 'int32_t

In the off chance that the v8 devs don't know this, <inttypes.h> (from C99) defines standard printf and scantf format specifiers for the core numeric types, e.g.

printf( "%"PRIi32"\n", myInt32);

Since i learned about those, all of my printf/scanf uses them, and my headaches with the above warnings went away. For my libraries which specify their own fixed-size number types, e.g.

typedef uint32_t mylib_id_t;

i also add:

#define MYLIB_ID_T_PFMT PRIu32

#define MYLIB_ID_T_SFMT SCNu32

--
----- stephan beal
http://wanderinghorse.net/home/stephan/

[pikpik]

Hi,

On Dec 14, 5:06 pm, Stephan Beal <sgb...@googlemail.com> wrote:
>
> In the off chance that the v8 devs don't know this, <inttypes.h> (from C99)
> defines standard printf and scantf format specifiers for the core numeric
> types, e.g.
>

[...]

Thanks Stephan.

Mind if I ask if this seems to be a general issue for V8, or does it
look like I did something wrong?

Thank you,
pikpik

[Kevin Millikin]

It's a V8 issue.  The code is changing all the time, so if you didn't change anything it's safe to assume V8 did :)

The argument types that don't agree with the format string seems like a bug.  I'll file an issue for you.  Here's the link to the V8 issue page (use "New Issue" to file) for future reference: http://code.google.com/p/v8/issues/list

--
v8-users mailing list
v8-u...@googlegroups.com
http://groups.google.com/group/v8-users

[pikpik]

Hi,

On Dec 17, 1:08 am, Kevin Millikin <kmilli...@chromium.org> wrote:
> It's a V8 issue.  The code is changing all the time, so if you didn't change
> anything it's safe to assume V8 did :)

Ok. Very cool. :)

> The argument types that don't agree with the format string seems like a bug.
>  I'll file an issue for you.  Here's the link to the V8 issue page (use "New
> Issue" to file) for future reference: http://code.google.com/p/v8/issues/list
>

Thanks!

pikpik