RE: host-ncurses 5.9 compiling fails on fedora 22 x86_64 distribution

[FengHaibo]

Hi Thomas & Mike:

The fix works to me.

Hi Micoslav:

Just wondering how release package in Fedora 22 passed compilation... using an older GCC?

Many thanks & Best Regards

Haibo

> Date: Mon, 8 Jun 2015 21:38:40 -0400
> From: address@hidden
> To: address@hidden
> CC: address@hidden; address@hidden
> Subject: Re: host-ncurses 5.9 compiling fails on fedora 22 x86_64 distribution
>
> On Mon, Jun 08, 2015 at 04:11:51PM +0000, Mike Gran wrote:
> > On Monday, June 8, 2015 5:17 AM, FengHaibo <address@hidden> wrote:
> >
> >
> >
> > >In fact I reported this issue to fedora release team and have no answer
> > >there, so looking help from ncurses, all other open source packages
> > >used are compiled well, only ncurses has the problem. Attached the
> > >generated lib_gen.c
>
> ncurses has a complicated configure script...
>
> > >As I checked, the error seems to be because automatically generated
> > >lib_gen.c has some syntax error like:
> > >NCURSES_EXPORT(int) (attroff) (int z)
> > >{
> > > T((T_CALLED("attroff(%s)"), _traceattr2(0,(chtype)z)));
> > >returnCode(wattr_off(stdscr, (attr_t)(z),
> > ># 39 "_19107.c" 3 4
> > > ((void *)0)
> > ># 39 "_19107.c"
> > >));
> > >}
> > >
> > >
> > >But in an old fedora version compiling, the line "# 39 "_19107.c""
> > >doesn't exist. This file is generated by a shell in ncurses, named
> > >MKlib_gen.sh
> > >
> > >
> > >Please kindly help to check this issue as it breaks my daily coding
> > Hi Thomas, et al. This looks like the porting problem
> > described here: https://gcc.gnu.org/gcc-5/porting_to.html
>
> This looks like the issue which (I think) was fixed in
>
> http://invisible-island.net/ncurses/NEWS.html#t20141206
>
> The fix would be to use a recent snapshot (with the caution about
> using --with-abi-version=5, if you want ncurses5, rather than ncurses6,
> of course):
>
> http://invisible-island.net/datafiles/current/ncurses.tar.gz
>
> > They say it is solved by adding the "-P" CFLAG to scripts that
> > automatically generate code.
>
> Actually that's not much of a solution - I try to avoid using compiler
> specific options to make code _portable_ (but it would have been nice if
> it was documented last summer when the problem was introduced).
>
> --
> Thomas E. Dickey <address@hidden>
> http://invisible-island.net
> ftp://invisible-island.net