3.0.4 release

[Benjamin Lindner]

Benjamin Lindner wrote:
> Jaroslav Hajek wrote:
>> hi,
>>
>> given the success of RC7, I declare the 3.0.4 release as
>> semi-officially released.
>> The tarballs are available from
>> http://artax.karlin.mff.cuni.cz/~hajej2am/ulozna/octave/
>>
>> I'll make an announcement to the help list when John uploads the
>> tarballs to GNU ftp sites.
>> John, would you, please?
>>
>> 3.0.4 is the last release of 3.0.x series, containing mostly bugfixes
>> and minor improvements over 3.0.3.
>>
>> enjoy
>>
> 
> I have built unger mingw (gcc-4.3.0-dw2-tdm) and I see a new failed test
> 
>   test_string.m .......................................... PASS 130/131  
> FAIL 1
> 
> Summary:
> 
>   PASS   4125
>   FAIL      1
> 
> The test log says:
> 
>   ***** test
>  charset = setstr (0:127);
> 
>  result = zeros (1, 128);
> 
>  result (33:127) = 1;
>  if (size(findstr (octave_config_info('canonical_host_type'), 'msvc'),2))
>    result(10) = 1;
>  endif
> 
>  assert(all (isprint (charset) == result));
> !!!!! test failed
> error: assert (all (isprint (charset) == result)) failed
> 
> 
> Now this one is mentioned in 
> http://www.nabble.com/3.0.4-RC6-to22703059.html
> 
> If I revert this patch, then all tests succed. If I apply it, then the 
> test fails.
> 
> What's the reason for this change?
> It would set a TAB character as printable ONLY for msvc builds, but not 
> for mingw builds.
> Why so?
> 
> #include <ctype.h>
> #include <stdio.h>
> 
> main() {
>    unsigned int i;
>    for( i=0; i<33; ++i ) {
>       printf("%d\tisprint=%d\tisspace=%d\tisblank=%d\tisalpha=%d\n",
>          i, isprint(i), isspace(i), isblank(i), isalpha(i));
>    }
>    return;
> }
> 
> I tested this on gcc-4.3.0-3-dw2-tdm and gcc-4.3.3-1-dw2-tdm and both
> return isblank(9)=true, so isprint(9) is true.
> (according to ctype.h a character is printable is its alphanumeric, 
> blank, punctuation or digit)
> 
> Tatsuro, do you see different behaviour here?
> 
> benjamin
> 

Sorry for the noise.
I just realised, that this patch does not change octave's behaviour (as 
I originally thought), it just changes the interpretation of a test result.
So I don't bother with it.

benjamin