9th January 2007 - 1.0.1 beta 21 ==================================

Download from

http://sourceforge.net/project/showfiles.php?group_id=141831&package_id=155759&release_id=477090


Summary:
--------

 * dynamic control of severity ceiling for fe.N
 * addition of missing implicit link file (be.file)
 * added fault-tolerance to be.N, so one or more (but not all) back-ends can
fail and the application can proceed
 * added be.fail back-end that always fails (for testing be.N fault
tolerance)
Details:
--------

Makefiles:

Application Layer:

Core:

Front-ends:

 + pantheios_fe_simple_getSeverityCeiling() and
   pantheios_fe_simple_setSeverityCeiling(), to dynamically
   control the severity ceiling

Back-ends:

 ~ fix minor syntax error troubling Borland, DMC++ and GCC
 + be.file.h
 ~ formatting error in back-ends on Win32 (fixed in STLSoft
   1.9.1 beta 39; download from http://stlsoft.org/downloads.html) + be.fail
 ~ be.N now tolerates failure of one or more (but not all) back-ends that
are flagged PANTHEIOS_BE_N_F_IGNORE_INIT_FAILURE
Examples:

Samples:

Depends on:

 * STLSoft 1.9.1 beta 40 or later
   (http://stlsoft.org/downloads.html)

gcc-4.1?

I copied build/gcc40 to build/gcc41 and tried it, I got:
STLSOFT=/usr/local/src/stlsoft make -f makefile.unix
g++   -c -Wall -pedantic -Wno-long-long -Wno-unused-value -Werror  -O1   -D_REENTRANT  -D_DEBUG -UNDEBUG  -Dunix -DPANTHEIOS_NO_AUTO_INIT  -I../../include -I"/usr/local/src/stlsoft/include"    -o pantheios_appl.debug.o ../../src/pantheios_appl.cpp
/usr/local/src/stlsoft/include/stlsoft/stlsoft.h: In destructor ‘stlsoft::stlsoft_size_checker::~stlsoft_size_checker()’:
/usr/local/src/stlsoft/include/stlsoft/stlsoft.h:1394: error: ISO C++ forbids zero-size array ‘ai’
make: *** [pantheios_appl.debug.o] Error 1

Neal Becker Wrote:

> gcc-4.1?
> 
> I copied build/gcc40 to build/gcc41 and tried it, I got:
> STLSOFT=/usr/local/src/stlsoft make -f makefile.unix
> g++   -c -Wall -pedantic -Wno-long-long -Wno-unused-value -Werror  -O1   -D_REENTRANT  -D_DEBUG -UNDEBUG  -Dunix -DPANTHEIOS_NO_AUTO_INIT  -I../../include -I"/usr/local/src/stlsoft/include"    -o pantheios_appl.debug.o ../../src/pantheios_appl.cpp
> /usr/local/src/stlsoft/include/stlsoft/stlsoft.h: In destructor ‘stlsoft::stlsoft_size_checker::~stlsoft_size_checker()’:
> /usr/local/src/stlsoft/include/stlsoft/stlsoft.h:1394: error: ISO C++ forbids zero-size array ‘ai’
> make: *** [pantheios_appl.debug.o] Error 1

That's very strange.

First, are you using STLSoft 1.9.1 b40?

If so, is line 1394:

        STLSOFT_STATIC_ASSERT(sizeof(ss_int32_t)  == 4);

I'm guessing a long on your architecture is not 32-bits, which is a reasonable thing, to be sure. What architecture are you running on?

Can you run the following, and let me know the results?

#include <stdio.h>

int main()
{
  printf("sizeof(short): %d\n", int(sizeof(short)));
  printf("sizeof(int): %d\n", int(sizeof(int)));
  printf("sizeof(long): %d\n", int(sizeof(long)));
  printf("sizeof(long long): %d\n", int(sizeof(long long)));

  return 0;
}

Thanks

Matthew Wilson wrote:

> Neal Becker Wrote:
> 
>> gcc-4.1?
>> 
>> I copied build/gcc40 to build/gcc41 and tried it, I got:
>> STLSOFT=/usr/local/src/stlsoft make -f makefile.unix
>> g++   -c -Wall -pedantic -Wno-long-long -Wno-unused-value -Werror  -O1
>> -D_REENTRANT  -D_DEBUG -UNDEBUG  -Dunix -DPANTHEIOS_NO_AUTO_INIT
>> -I../../include -I"/usr/local/src/stlsoft/include"    -o
>> pantheios_appl.debug.o ../../src/pantheios_appl.cpp
>> /usr/local/src/stlsoft/include/stlsoft/stlsoft.h: In destructor
>> ‘stlsoft::stlsoft_size_checker::~stlsoft_size_checker()’:
>> /usr/local/src/stlsoft/include/stlsoft/stlsoft.h:1394: error: ISO C++
>> forbids zero-size array ‘ai’ make: *** [pantheios_appl.debug.o] Error 1
> 
> That's very strange.
> 
> First, are you using STLSoft 1.9.1 b40?

Yes.

> 
> If so, is line 1394:
> 
>         STLSOFT_STATIC_ASSERT(sizeof(ss_int32_t)  == 4);
Yes.


> 
> I'm guessing a long on your architecture is not 32-bits, which is a reasonable thing, to be sure. What architecture are you running on?

This is linux, fedora fc6 on x86_64.

> 
> Can you run the following, and let me know the results?
> 
> #include <stdio.h>
> 
> int main()
> {
>   printf("sizeof(short): %d\n", int(sizeof(short)));
>   printf("sizeof(int): %d\n", int(sizeof(int)));
>   printf("sizeof(long): %d\n", int(sizeof(long)));
>   printf("sizeof(long long): %d\n", int(sizeof(long long)));
> 
>   return 0;
> }
> 
> Thanks
./Test
sizeof(short): 2
sizeof(int): 4
sizeof(long): 8
sizeof(long long): 8

Matthew Wilson wrote:

> Neal Becker Wrote:
> 
>> gcc-4.1?
>> 
>> I copied build/gcc40 to build/gcc41 and tried it, I got:
>> STLSOFT=/usr/local/src/stlsoft make -f makefile.unix
>> g++   -c -Wall -pedantic -Wno-long-long -Wno-unused-value -Werror  -O1
>> -D_REENTRANT  -D_DEBUG -UNDEBUG  -Dunix -DPANTHEIOS_NO_AUTO_INIT
>> -I../../include -I"/usr/local/src/stlsoft/include"    -o
>> pantheios_appl.debug.o ../../src/pantheios_appl.cpp
>> /usr/local/src/stlsoft/include/stlsoft/stlsoft.h: In destructor
>> ‘stlsoft::stlsoft_size_checker::~stlsoft_size_checker()’:
>> /usr/local/src/stlsoft/include/stlsoft/stlsoft.h:1394: error: ISO C++
>> forbids zero-size array ‘ai’ make: *** [pantheios_appl.debug.o] Error 1
> 
> That's very strange.
> 
> First, are you using STLSoft 1.9.1 b40?
> 
> If so, is line 1394:
> 
>         STLSOFT_STATIC_ASSERT(sizeof(ss_int32_t)  == 4);
> 
> I'm guessing a long on your architecture is not 32-bits, which is a reasonable thing, to be sure. What architecture are you running on?
> 
> Can you run the following, and let me know the results?
> 
> #include <stdio.h>
> 
> int main()
> {
>   printf("sizeof(short): %d\n", int(sizeof(short)));
>   printf("sizeof(int): %d\n", int(sizeof(int)));
>   printf("sizeof(long): %d\n", int(sizeof(long)));
>   printf("sizeof(long long): %d\n", int(sizeof(long long)));
> 
>   return 0;
> }
> 
> Thanks

Here's what I see in preprocessed output:

Error occurs here:
        do { typedef int ai[(sizeof(ss_int32_t) == 4) ? 1 : 0]; } while(0);

What is ss_int32_t?
typedef ss_int32_pr_t_ ss_int32_t;
# 1218 "/usr/local/src/stlsoft/include/stlsoft/stlsoft.h"
 typedef long ss_int32_pr_t_;

"Neal Becker" <ndbecker2@gmail.com> wrote in message news:eo9453$1nf4$1@digitaldaemon.com...
> Matthew Wilson wrote:
>
> > Neal Becker Wrote:
> >
> >> gcc-4.1?
> >>
> >> I copied build/gcc40 to build/gcc41 and tried it, I got:
> >> STLSOFT=/usr/local/src/stlsoft make -f makefile.unix
> >> g++   -c -Wall -pedantic -Wno-long-long -Wno-unused-value -Werror  -O1
> >> -D_REENTRANT  -D_DEBUG -UNDEBUG  -Dunix -DPANTHEIOS_NO_AUTO_INIT
> >> -I../../include -I"/usr/local/src/stlsoft/include"    -o
> >> pantheios_appl.debug.o ../../src/pantheios_appl.cpp
> >> /usr/local/src/stlsoft/include/stlsoft/stlsoft.h: In destructor
> >> 'stlsoft::stlsoft_size_checker::~stlsoft_size_checker()':
> >> /usr/local/src/stlsoft/include/stlsoft/stlsoft.h:1394: error: ISO C++
> >> forbids zero-size array 'ai' make: *** [pantheios_appl.debug.o] Error 1
> >
> > That's very strange.
> >
> > First, are you using STLSoft 1.9.1 b40?
>
> Yes.
>
> >
> > If so, is line 1394:
> >
> >         STLSOFT_STATIC_ASSERT(sizeof(ss_int32_t)  == 4);
> Yes.
>
>
> >
> > I'm guessing a long on your architecture is not 32-bits, which is a reasonable thing, to be sure. What architecture are you running on?
>
> This is linux, fedora fc6 on x86_64.
>
> >
> > Can you run the following, and let me know the results?
> >
> > #include <stdio.h>
> >
> > int main()
> > {
> >   printf("sizeof(short): %d\n", int(sizeof(short)));
> >   printf("sizeof(int): %d\n", int(sizeof(int)));
> >   printf("sizeof(long): %d\n", int(sizeof(long)));
> >   printf("sizeof(long long): %d\n", int(sizeof(long long)));
> >
> >   return 0;
> > }
> >
> > Thanks
> ./Test
> sizeof(short): 2
> sizeof(int): 4
> sizeof(long): 8
> sizeof(long long): 8

Cool.

Your post has stimulated some long-put-off refactoring of the type discrimination. I've been working all morning on this, and am pretty happy that it's all to the good.

I'm now changing the underlying 32-bit types to be int wherever appropriate, to avoid the issue you've discovered.

Should be able to release b41 later today. Assuming that goes well, I'm but a hair's breadth away from releasing 1.9.1. (It's only been 16 months! <g>)

Cheers

Matthew

> Here's what I see in preprocessed output:
> 
> Error occurs here:
>         do { typedef int ai[(sizeof(ss_int32_t) == 4) ? 1 : 0]; } while(0);
> 
> What is ss_int32_t?

It's a 32-bit integer type. Equivalent to C99's int32_t.

> typedef ss_int32_pr_t_ ss_int32_t;
> # 1218 "/usr/local/src/stlsoft/include/stlsoft/stlsoft.h"
>  typedef long ss_int32_pr_t_;

This is the internal types guff. I've now refactored this significantly, as you'll see with b 41

Cheers

Matthew

Matthew Wilson wrote:

>
> 
> Should be able to release b41 later today. Assuming that goes well, I'm but a hair's breadth away from releasing 1.9.1. (It's only been 16 months! <g>)
> 
> Cheers
> 

Great, I'm looking forward to it.

BTW, how about this: http://boost.org/libs/integer/integer.htm#sized