Re: [Help-gsl] Building gsl-1.15 under MinGW

[Sisyphus]

----- Original Message ----- 
From: "Armin Armbruster" <address@hidden>
To: <address@hidden>; "Sisyphus" <address@hidden>
Sent: Thursday, August 18, 2011 4:02 AM
Subject: Re: [Help-gsl] Building gsl-1.15 under MinGW


Hi Rob,

Thanks for your reply.

here's gcc -v:

$ gcc -v
Using built-in specs.
COLLECT_GCC=C:\MinGW\bin\gcc.exe
COLLECT_LTO_WRAPPER=c:/mingw/bin/../libexec/gcc/mingw32/4.5.2/lto-wrapper.exe
Target: mingw32
Configured with: 
../gcc-4.5.2/configure --enable-languages=c,c++,ada,fortran,objc,obj-c++ --disable-sjlj-exceptions 
--wi
th-dwarf2 --enable-shared --enable-libgomp --disable-win32-registry --enable-libstdcxx-debug 
--enable-version-specific-r
untime-libs --disable-werror --build=mingw32 --prefix=/mingw
Thread model: win32
gcc version 4.5.2 (GCC)


Some more background:
- I'm using the latest MinGW (downloaded from 
http://sourceforge.net/projects/mingw/files/Automated%20MinGW%20Installer/mingw-get-inst/), 
mingw-get-inst-20110802.exe).
- $ help ==>GNU bash, version 3.1.17(1)-release (i686-pc-msys)
- my host OS is MS Windows XP, SP3
- gsl downloaded from ftp://ftp.gnu.org/gnu/gsl/gsl-1.15.tar.gz

Last night I've tried to build the same gsl on my computer at home (running 
Windows 7 Home Premium, SP1, 64 bit OS). I used the same MinGW installer and 
the same GSL package and make ran without problems.

I compared the config.log files in both cases (see attachment 
config_log_diff.txt). The only significant difference that I can see is that 
uname -s returns MINGW32_NT-5.1 on the XP machine and MINGW32_NT-6.1 on the 
Win7 machine.

The one thing I noticed is, that the resulting config.h is a direct copy of 
config.h.in!!! The only difference is the first line added to config.h. That 
makes me believe that something with awk and/or sed isn't working properly. 
I don't have a full understanding of the configure script but from what I 
understand it is taking the config.h.in file as a template and mangles it 
based on some test results to turn on some of the defines. Is that right?

=============================

I don't understand it all that well either, but I believe your understanding 
is basically correct.

In addition to the gcc-3.4.5 that I used yesterday, I also have the very 
same setup that you have - namely the one installed by 
mingw-get-inst-20110802.exe.

When I use that, a correct config.h is still being created for me

I'm fairly sure that your only problem is that your config.h is not being 
generated correctly ... and I have no idea why that happens.

The only differences in our config.log files is the things you would expect 
to see as the result of having been run on different machines, different 
pathnames, etc.
I can see nothing that explains the reason that your config.h turns out 
botched.

Comparing our 'uname' info:

uname -m = i686
uname -r = 1.0.17(0.48/3/2)
-uname -s = MINGW32_NT-5.1
+uname -s = MINGW32_NT-6.0
uname -v = 2011-04-24 23:39

we see that everything is exactly the same, except for 'uname -s'. I presume 
that's merely a reflection of different operating systems (XP versus Vista).

We probably need some assistance from someone with a better understanding of 
how this config file is supposed to be generated.

Cheers,
Rob