Re: RPM specfile error with F30

[Phil Wyett]

On Fri, 2019-05-31 at 20:32 +0100, Phil Wyett wrote:
> On Fri, 2019-05-31 at 13:24 -0500, Ranjan Maitra wrote:
> > On Fri, 31 May 2019 19:58:04 +0200 Franta Hanzlík <
> > [e-mail]
> > > wrote:
> > > On Wed, 29 May 2019 18:06:19 -0500
> > > Ranjan Maitra  wrote:
> > > > Hi,
> > > >
> > > > I have the following specfile which compiled fine in F29 but is
> > > > not able to find references to standard functions in X11, etc
> > > > even though the compilation happens without error (outside the
> > > > rpmbuild environment). So I am wondering if anyone has some
> > > > suggestions on what has changed/is going wrong.
> > > >
> > > > Here is the spec file:
> > > >
> > > > $ fpaste thaali.spec
> > > > Uploading (1.4KiB)...
> > > > https://paste.fedoraproject.org/paste/3sHHnOdGoM4YHmdzzJRzgQ
> > > >
> > > > And here are the messages that I get when I try to build:
> > >
> > > ...
> > >
> > > Just an idea - once I had a problem building a program when the
> > > make
> > > '-j' switch was greater than one. Specifying '-j1' solved it.
> > > Unfortunately, I don't even remember the cause or how it
> > > manifested
> > > itself...
> >
> > Thanks! So, I use:
> >
> > %make_build
> >
> > I replaced this with "make -j1" or also with "make -j" or -j or -j1
> > appended to %make_build.
> >
> > I don't really understand this, but all of them gave errors.
> >
> > Thanks again for your help!
> >
> > Best wishes,
> > Ranjan
> > _______________________________________________
>
> Hi Ranjan,
>
> This is an upstream configure issue.
> I took the pastebin spec file.
> Added to spec file: BuildRequires: gcc
> Saw spec file does a reconfigure and looked at upstream
>
> configure.ac
> 4. Saw it is not correct. Substitute upstream configure.ac for below.
>
> // Start
>
> AC_INIT(thaali, 0.4.0)
> AC_CONFIG_SRCDIR(src/thaali.c)
> AM_INIT_AUTOMAKE
>
> AM_CONFIG_HEADER(config.h)
>
> AC_PROG_CC
> AC_PATH_X
> AC_PATH_XTRA
>
> dnl Are we going to enable the debugging output ??
> AC_ARG_ENABLE(debug,
> AC_HELP_STRING([--enable-debug], [Enable runtime debugging (default:
> no)]),
>         debug=$enableval, debug=no)
>
> if test "$debug" = yes; then
>         AC_DEFINE(DEBUG,,[Enable verbose debugging output])
> fi
>
> dnl Let's do some X testing for... ;-)
> if test -n "$x_includes" && test "x$x_includes" != xNONE ; then
>   CFLAGS="$CFLAGS -I$x_includes"
> fi
> if test -n "$x_libraries" && test "x$x_libraries" != xNONE ; then
>   LDFLAGS="$LDFLAGS -L$x_libraries"
> fi
>
> AC_CHECK_LIB([ICE],[IceOpenConnection],,AC_MSG_ERROR([Can't find
> libICE]),)
> AC_CHECK_LIB([SM],[SmcOpenConnection],,AC_MSG_ERROR([Can't find
> libSM]),)
> AC_CHECK_LIB([X11],[XOpenDisplay],,AC_MSG_ERROR([Can't find libX11]),
>             -L$x_libraries $X_PRE_LIBS)
>
> AC_SUBST(x_includes)
> AC_SUBST(x_libraries)
>
> AC_OUTPUT([
> Makefile
> src/Makefile
> ])
>
> // End
>
> Now the RPM builds fine on F30.
>
> Note this was just a hack to get it to build and show it as an
> upstream
> issue. You can diff my changes against original configure.ac. To be
> honest you have some work ahead as this is all very old and out of
> date
> both upstream side and how the spec file is constructed. You will
> need
> to research to fix.
>
> Regards
>
> Phil

Hi,

For such a small application, you would vastly simplify things with a
build system like meson.

Regards

Phil