win32utils devel - Aug 2009 - win32-api, unsigned long vs long in api_call

Hi,

I noticed this bug:

Windows XP Pro
Ruby 1.8.6-p383 built with VC++ 9
win32-api 1.4.3

C:\>irb
irb(main):001:0> require ''win32/api''
=> true
irb(main):002:0> GetFileAttributes
Win32::API.new(''GetFileAttributes'', ''S'',
''L'')
=> #<Win32::API:0x2db8880>
irb(main):003:0> GetFileAttributes.call("C:/aaabbbcccxxx") # bogus
file, should return -1
=> 4294967295

I looked at win32-api.c, line 743, and I saw that, in the api_call
function, we''ve declared the return_value variable an unsigned long. I
changed it to a signed long as well as switching a ULONG2NUM call to
LONG2NUM and that fixed it.

I''m kinda shocked I didn''t see this until now. Is there any
reason for
me not to apply this change immediately? I''m trying to remember why I
made it unsigned to begin with.

Regards,

Dan

On Sun, Aug 16, 2009 at 2:14 PM, Daniel Berger<djberg96 at gmail.com>
wrote:
> Hi,
>
> I noticed this bug:
>
> Windows XP Pro
> Ruby 1.8.6-p383 built with VC++ 9
> win32-api 1.4.3
>
> C:\>irb
> irb(main):001:0> require ''win32/api''
> => true
> irb(main):002:0> GetFileAttributes >
Win32::API.new(''GetFileAttributes'', ''S'',
''L'')
> => #<Win32::API:0x2db8880>
> irb(main):003:0> GetFileAttributes.call("C:/aaabbbcccxxx") #
bogus
> file, should return -1
> => 4294967295
>
> I looked at win32-api.c, line 743, and I saw that, in the api_call
> function, we''ve declared the return_value variable an unsigned
long. I
> changed it to a signed long as well as switching a ULONG2NUM call to
> LONG2NUM and that fixed it.
>
> I''m kinda shocked I didn''t see this until now. Is there
any reason for
> me not to apply this change immediately? I''m trying to remember
why I
> made it unsigned to begin with.
I''ve gone ahead and changed it for 1.4.4. If it turns out to be a
critical mistake I''ll undo it.

Regards,

Dan