Herbert,
Thanks for the information. I attached the new mathconf.h file with your patch.
I then did a clean build and it compiles without error. I am not sure about
what you'd like me to provide when you say , " Of course it would also
interesting to know what type ::std::numeric_limits<unsigned char>::digits on
that platform." Could you provide more details? Thanks.
Also,
I've found a couple other things that you may be interested in.
1. The generated bootstrap script will not execute after configure because of
like 64. I've had to change it like this:
- dmake_full_package_name=$(find $TARFILE_LOCATION -type f
-name "*-$dmake_package_name")
+ dmake_full_package_name=`find $TARFILE_LOCATION -type f
-name "*-$dmake_package_name"`
2. I keep receiving the following each time I do a clean build:
=============
Building module basebmp
=============
Entering /opt/aoo-4.0.0/main/basebmp/source
Compiling: basebmp/source/bitmapdevice.cxx
"/opt/aoo-4.0.0/main/solver/400/unxsoli4.pro/inc/vigra/mathutil.hxx", line 801:
Error: erf is not a member of _STL.
1 Error(s) detected.
dmake: Error code 2, while making '../unxsoli4.pro/slo/bitmapdevice.obj'
1 module(s):
basebmp
need(s) to be rebuilt
Reason(s):
ERROR: error 65280 occurred while making /opt/aoo-4.0.0/main/basebmp/source
When you have fixed the errors in that module you can resume the build by
running:
build --all:basebmp
I can resolve this by changing
/opt/aoo-4.0.0/main/solver/400/unxsoli4.pro/inc/vigra/mathutil.hxx to the
following on line 801:
- using VIGRA_CSTD::erf;
+ using ::erf;
This file is generated by the build. As of now, I have not determined what is
generating it.
-----Original Message-----
From: Herbert Duerr [mailto:[email protected]]
Sent: Monday, September 09, 2013 12:58 AM
To: [email protected]
Cc: Steele, Raymond
Subject: Re: EXTERNAL: Re: Building comphelper
On 07.09.2013 00:17, Steele, Raymond wrote:
> To address the isfinite() issue I added the following to rtl/math.hxx:
>
> #ifdef SOLARIS
> #undef SAL_MATH_FINITE
> #define SAL_MATH_FINITE(d) finite(d)
> #endf
>
> Your links were very helpful. Thanks so much.
Wonderful, so this problem is mostly solved. We just need to get it cleanly
into the mathconf.h. How about this patch?
--- main/sal/inc/sal/mathconf.h
+++ main/sal/inc/sal/mathconf.h
@@ -55,7 +55,9 @@ extern "C" {
/* SAL_MATH_FINITE(d): test double d on INFINITY, NaN et al. */ -#if
defined(__GNUC__) // workaround gcc bug 14608
+#if defined(OS2) || defined(SOLARIS)
+ #define SAL_MATH_FINITE(d) finite(d)
+#elif defined(__GNUC__) // workaround gcc bug 14608
#if (__GNUC_MINOR__ >= 3) // gcc>=4.3 has a builtin
#define SAL_MATH_FINITE(d) __builtin_isfinite(d)
#else
@@ -67,8 +69,6 @@ extern "C" {
#define SAL_MATH_FINITE(d) isfinite(d)
#elif defined( WNT)
#define SAL_MATH_FINITE(d) _finite(d)
-#elif defined OS2
-#define SAL_MATH_FINITE(d) finite(d)
#elif defined LINUX || defined UNX
#define SAL_MATH_FINITE(d) finite(d)
#else /* WNT, LINUX, UNX */
> Now onto:
>
> integer_fwd.cpp contains the following on line 136 and 137:
>
> 136: template < > 137: struct low_bits_mask_t<
> ::std::numeric_limits<unsigned
> char>::digits>;
I'm not sure whether you saw my mail from September 2nd
(http://markmail.org/message/bt7bazzvb6zionee):
https://issues.apache.org/jira/browse/STDCXX-937 might have to do with this. It
mentions a compiler bug 6703971 but that doesn't seem to be available any more.
I suggest to update the compiler to its latest patch level.
Of course it would also interesting to know what type
::std::numeric_limits<unsigned char>::digits on that platform. The error
message looks as if this was not an integer but a class. The C++ spec requires
that the type should be an int. Please check the <limits> include file.
Herbert
/**************************************************************
*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*
*************************************************************/
#if !defined INCLUDED_SAL_MATHCONF_H
#define INCLUDED_SAL_MATHCONF_H
#include "osl/endian.h"
#include <float.h>
#include <math.h>
#if defined SOLARIS
#include <ieeefp.h>
#endif /* SOLARIS */
#if defined __cplusplus
extern "C" {
#endif /* __cplusplus */
/* Generally, the C standard guarantees that at program startup, "trapping or
stopping (if supported) is disabled on all [floating-point] exceptions"
(F.7.3/1 of the August 3, 1998 draft of C99), and that during program
execution, "a programmer can safely assume default modes (or be unaware of
them)" (7.6/2, footnote 161 of the August 3, 1998 draft of C99). Reportedly,
on Windows there are printer drivers that switch on exceptions. To avoid
problems, the SAL_MATH_FPEXCEPTIONS_OFF macro can be used to explicitly
switch off exceptions (on Windows).
*/
#if defined WNT
#define SAL_MATH_FPEXCEPTIONS_OFF() _control87( _MCW_EM, _MCW_EM )
#else /* WNT */
#define SAL_MATH_FPEXCEPTIONS_OFF()
#endif /* WNT */
/* SAL_MATH_FINITE(d): test double d on INFINITY, NaN et al. */
#if defined(OS2) || defined(SOLARIS)
#define SAL_MATH_FINITE(d) finite(d)
#elif defined(__GNUC__) // workaround gcc bug 14608
#if (__GNUC_MINOR__ >= 3) // gcc>=4.3 has a builtin
#define SAL_MATH_FINITE(d) __builtin_isfinite(d)
#else
#define SAL_MATH_FINITE(d) finite(d) // fall back to pre-C99
name
#endif
#elif defined(__STDC__)
// isfinite() should be available in math.h according to
C99,C++99,SUSv3,etc.
// unless GCC bug 14608 hits us where cmath undefines isfinite() as
macro
#define SAL_MATH_FINITE(d) isfinite(d)
#elif defined( WNT)
#define SAL_MATH_FINITE(d) _finite(d)
#elif defined LINUX || defined UNX
#define SAL_MATH_FINITE(d) finite(d)
#else /* WNT, LINUX, UNX */
#error "SAL_MATH_FINITE not defined"
#endif /* WNT, LINUX, UNX */
/* This needs to be fixed for non--IEEE-754 platforms: */
#if 1 /* IEEE 754 supported */
#if defined OSL_BIGENDIAN
/* IEEE 754 double structures for BigEndian */
union sal_math_Double
{
struct
{
unsigned sign : 1;
unsigned exponent :11;
unsigned fraction_hi :20;
unsigned fraction_lo :32;
} inf_parts;
struct
{
unsigned sign : 1;
unsigned exponent :11;
unsigned qnan_bit : 1;
unsigned bits :19;
unsigned fraction_lo :32;
} nan_parts;
struct
{
unsigned msw :32;
unsigned lsw :32;
} w32_parts;
double value;
};
#elif defined OSL_LITENDIAN
/* IEEE 754 double structures for LittleEndian */
union sal_math_Double
{
struct {
unsigned fraction_lo :32;
unsigned fraction_hi :20;
unsigned exponent :11;
unsigned sign : 1;
} inf_parts;
struct {
unsigned fraction_lo :32;
unsigned bits :19;
unsigned qnan_bit : 1;
unsigned exponent :11;
unsigned sign : 1;
} nan_parts;
struct
{
unsigned lsw :32;
unsigned msw :32;
} w32_parts;
double value;
};
#else /* OSL_BIGENDIAN, OSL_LITENDIAN */
#error "neither OSL_BIGENDIAN nor OSL_LITENDIAN"
#endif /* OSL_BIGENDIAN, OSL_LITENDIAN */
#else /* IEEE 754 supported */
#error "don't know how to handle IEEE 754"
#endif /* IEEE 754 supported */
#if defined __cplusplus
}
#endif /* __cplusplus */
#endif /* INCLUDED_SAL_MATHCONF_H */
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
