FW: FW: [Help-glpk] bigmul in glblib05.c

[Cheng,Jen-Min]

These codes are replaced with 

unsigned short c = (unsigned short)( i & USHRT_MAX);

The modified codes can produce 13.75 for vpm2.mps of MIPLIB 2003.

Thanks.  


-----Original Message-----
From: Michael Hennebry [mailto:address@hidden 
Sent: Tuesday, January 05, 2010 10:44 AM
To: Andrew Makhorin
Cc: Cheng,Jen-Min; address@hidden
Subject: Re: FW: [Help-glpk] bigmul in glblib05.c

On Tue, 5 Jan 2010, Andrew Makhorin wrote:

>> Run-Time Check Failure #1 - A cast to a smaller data type has
>> caused a loss of data. If this was intentional, you should mask
>> the source of the cast with the appropriate bitmask.  For
>> example:
>>
>> char c = (i & 0xFF);

What this does is implementation dependent.

char c = (char)i;
char c = (char)(i & 0xFF);
char c = (char)(i & CHAR_MAX);
char c = (char)(i & UCHAR_MAX);
char c = (char)(unsigned char)i;
would all be better, but all still implementation dependent.
None should give you a warning.
They might give you five different results.
If chars are unsigned,
all except possibly the second will do what you probably want.
Life is more interesting if chars are signed.
C defines conversions from unsigned to
signed types only in their common range.
The others are implemenation defined.

> This advice is helpful for quiche eaters only :)
>
>> Because MSVS 2008 would issue warning for such a conversion, the
bigmul
>> is modified as
>
>>       union {
>>          unsigned int uint;
>>          struct { unsigned short ush1, ush2 ;} ush;
>>       } tuiush;
>> [...]
>>                x[i+j] = (unsigned short)tuiush.ush.ush1;

If preceded by an assignment to tuish.uint, it would be undefined
behaviour.

-- 
Michael   address@hidden
"Pessimist: The glass is half empty.
Optimist:   The glass is half full.
Engineer:   The glass is twice as big as it needs to be."