|
From: | Wistar, Jonathan |
Subject: | [Bug-tar] GNU tar V1.15.1: "sysexits.h: No such file or directory" |
Date: | Thu, 27 Apr 2006 12:37:49 -0400 |
Hello, Likely this has been reported before, but I was
unsure what to consult in order to confirm or deny that theory. On the
off-chance that this bug has not already been reported, here are the details… My system type is SFU (a.k.a. Software For Unix
= Interix) V3.5. I was attempting to build GNU tar V1.15.1. It appears that "configure" correctly diagnosed
that this system has no "sysexists.h" header file. The output
from "configure" is: checking
sysexits.h usability… no checking
sysexits.h presence… no checking
for sysexits.h… no and the details in "config.log" are: configure:19425: checking sysexits.h
usability conftest.c:174:22: sysexits.h: No
such file or directory [
a bit later ] | #include <sysexits.h> configure:19470: checking sysexits.h
presence conftest.c:140:22 sysexits.h: No
such file or directory [
a bit later ] | #include <sysexits.h> configure:19541: checking for
sysexits.h [
a bit later ] ac_cv_header_sysexits_h=no The top-level (generated) "config.h" shows: /*
Define to 1 if you have the <sysexits.h> header file. */ /*
#undef HAVE_SYSEXITS_H */ which is clearly correct. The top-level (generated) "Makefile", the macro
"am__aclocal_m4_deps" has $(top_srcdir)/m4/sysexits.m4 in its list. (Presumably,
this is okay.) So, "configure" completes without an error. But, when I next do: %
make eventually, it attempts to compile "argp-eexst.c"
(in the "lib" sub-directory): if
gcc -DHAVE_CONFIG_H -DLIBDIR=\"/usr/local/lib\" -I. - argp-eexst.o
-MD -MP -MF ".deps/argp-eexst.Tpo" -c -o argp-eexst.o \ argp-eexst.c;
then mv -f "./deps/argp--eexst.Tpo" ".deps/argp-eexst.Po";\ else
rm -f ".deps/argp-eexst.Tpo"; exit 1; fi which produces: argp-eexst.c:24:22:
sysexits.h: No such file or directory argp-eexst.c:31:
error: `EX_USAGE' undeclared here (not in a function) ***
Error code 1 Stop. [etc.,
etc.] And, of course, the source code for the program "argp-eexst.c"
looks like: #ifdef
HAVE_CONFIG_H #include
<config.h> #endif #include
<sysexits.h> #include
"argp.h" [etc.] So, given that I find no "sysexits.h" inside this
GNU package (despite the "m4" file "sysexits.m4" file), it seems that the
bug is not a missing "-I" argument to the compiler but rather a failure to utilize properly the information
already collected by "configure" (in the "config.h" header file) regarding the
lack of a system header file "sysexits.h". This does not appear to be a problem specific to my
system type (SFU). Regards, J Wistar |
[Prev in Thread] | Current Thread | [Next in Thread] |