Compiling GMP 5.0.5 - Printable Version

Compiling GMP 5.0.5 - Printable Version

+- IRIX Network Forums (//forums.irixnet.org)
+-- Forum: SGI/MIPS (//forums.irixnet.org/forum-3.html)
+--- Forum: IRIX/Software (//forums.irixnet.org/forum-7.html)
+--- Thread: Compiling GMP 5.0.5 (/thread-198.html)

Compiling GMP 5.0.5 - LarBob - 07-16-2018

Hi, I'm trying to compile GMP 5.0.5 with GCC 4.3.1 and am running into a weird issue that I haven't run into before when compiling it on IRIX.

Code:
../gmp-impl.h:189: error: expected '=', ',', ';', 'asm' or '__attribute__' before 'gmp_uint_least32_t'

There's not an issue with the gmp-impl header obviously, but I'm not exactly sure where the problem actually lies. Any suggestions?

Dev environment info:

Code:
versions | grep compiler.dev

I  compiler_dev         07/16/2018  Base Compiler Development Environment, 7.4

I  compiler_dev.books   07/16/2018  Base Compiler Books

I  compiler_dev.books.MProAsLg_PG  07/16/2018  MIPSpro Assembly Language Programming Guide

I  compiler_dev.books.MProCplrDbx_TG  07/16/2018  MIPSpro N32/64 Compiling and Performance Tuning Guide

I  compiler_dev.books.MPro_n32_ABI  07/16/2018  MIPSpro N32 ABI Handbook

I  compiler_dev.books.Pragmas  07/16/2018  MIPSpro C/C++ Pragmas

I  compiler_dev.hdr     07/16/2018  Base Compiler Headers

I  compiler_dev.hdr.lib 07/16/2018  Base Compiler Headers

I  compiler_dev.man     07/16/2018  Base Compiler Manual Pages

I  compiler_dev.man.base  07/16/2018  Base Compiler Components Man Pages

I  compiler_dev.man.ld  07/16/2018  Base Compiler Loader Man Pages

I  compiler_dev.man.lib 07/16/2018  Base Compiler Programmers Man Pages

I  compiler_dev.man.relnotes  07/16/2018  Base Compiler Release Notes

I  compiler_dev.man.util  07/16/2018  Base Compiler Utility Man Pages

I  compiler_dev.sw      07/16/2018  Base Compiler Software

I  compiler_dev.sw.base 07/16/2018  Base Compiler Components

I  compiler_dev.sw.ld   07/16/2018  Base Compiler Loader

I  compiler_dev.sw.lib  07/16/2018  Base Compiler Libraries

I  compiler_dev.sw.util 07/16/2018  Base Compiler Utilities

I  compiler_dev.sw64    07/16/2018  Base Compiler Software

I  compiler_dev.sw64.lib  07/16/2018  Base Compiler Libraries 64bit

I  compiler_dev         07/15/2018  Base Compiler Development Environment, 7.3

I  compiler_dev.books   07/15/2018  Base Compiler Books

I  compiler_dev.books.MProPort_TG  07/15/2018  MIPSpro Porting and Transition Guide

Re: Compiling GMP 5.0.5 - jpstewart - 07-16-2018

Line 189 of gmp-impl.h (as mentioned in the error message) tries to typedef gmp_uint_least32_t to the operating system's uint_least32_t type. On IRIX that's defined in stdint.h but only for C99 mode. So if you're compiling C++ there's no uint_least32_t available.

Also, gmp-impl.h will prefer to use inttypes.h instead of stdint.h (lines 110 to 116) if both are available. On IRIX, inttypes.h doesn't define uint_least32_t at all, regardless of mode.

My solution (when compiling gmp-4.3.2 a few years ago) was to just pass ac_cv_type_uint_least32_t=no to configure before building GMP to force it to fall back to other code (gmp-impl.h lines 191 to 200). Fortunately, GMP is well-written in that regard! That work-around should also suffice for the version you're trying.

Re: Compiling GMP 5.0.5 - LarBob - 07-16-2018

<QUOTE author="jpstewart" post_id="1543" time="1531755020" user_id="129">

jpstewart post_id=1543 time=1531755020 user_id=129 Wrote:Line 189 of gmp-impl.h (as mentioned in the error message) tries to typedef gmp_uint_least32_t to the operating system's uint_least32_t type. On IRIX that's defined in stdint.h but only for C99 mode. So if you're compiling C++ there's no uint_least32_t available.

Also, gmp-impl.h will prefer to use inttypes.h instead of stdint.h (lines 110 to 116) if both are available. On IRIX, inttypes.h doesn't define uint_least32_t at all, regardless of mode.

My solution (when compiling gmp-4.3.2 a few years ago) was to just pass ac_cv_type_uint_least32_t=no to configure before building GMP to force it to fall back to other code (gmp-impl.h lines 191 to 200). Fortunately, GMP is well-written in that regard! That work-around should also suffice for the version you're trying.

Hm, thank you, I'll try that. What's weird is that I didn't have to do that before when compiling it on another machine though...