bug-texinfo
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [platform-testers] texinfo-6.5.90 pretest on Solaris 10


From: Gavin Smith
Subject: Re: [platform-testers] texinfo-6.5.90 pretest on Solaris 10
Date: Sat, 5 Jan 2019 20:36:19 +0000

On 1/5/19, Bruno Haible <address@hidden> wrote:
> In the config.log of the tp/Texinfo/XS directory:
>
> configure:7518: cc -o conftest -g -I/home/haible/prefix-x86_64/include
> -L/home/haible/prefix-x86_64/lib conftest.c
> /home/haible/prefix-x86_64/lib/libiconv.so -R/home/haible/prefix-x86_64/lib
>>&5
> ld: fatal: file /home/haible/prefix-x86_64/lib/libiconv.so: wrong ELF class:
> ELFCLASS64
> ld: fatal: file processing errors. No output written to conftest
> configure:7518: $? = 1
>
> As you can see, it does not use $CC (which I set to a 64-bit compiler)
> but 'cc' (which happens to be a 32-bit compiler - thus causing a mismatch
> with the CPPFLAGS and LDFLAGS that I specified).

The CPPFLAGS and LDFLAGS do need to be removed, but there is another
problem. Even with these removed, if the configure script sets the
"HAVE_ICONV" preprocessor symbol to 0, then iconv.h won't be included
and iconv_t won't be defined. I found this would happen on Solaris 10
due to this bug, as documented in the gnulib manual:

> Failures are not distinguishable from successful returns on some
>     platforms: AIX 5.1, Solaris 10.

It doesn't appear that the "iconv" gnulib module actually fixes this
problem, although in the manual it says it does.

Probably what we'll have to do (in addition to sorting out the problem
with CPPFLAGS) is make building Parsetexi.la subject to an automake
conditional that depends on whether iconv was found to work.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]