[GRASS-dev] compile errors when using stripped-down config

I’m working on packaging 7.0.1 (well, writing a new ebuild) for Gentoo Linux. When testing a totally stripped down configuration (ie, everything disabled that could be), I ran into a couple of compile errors. Here are things that need to be fixed:

#1: ./lib/raster/open.c needs to include <errno.h> explicitly. Otherwise errno is undefined at line 595.

#2: ./general/g.proj/main.c declares the variable “inepsg” only if HAVE_OGR is defined, but then uses this symbol outside “#ifdef HAVE_OGR” in lines 302 and 322. I’m not exactly sure how to fix this.


Soren Harward
stharward@gmail.com

Soren Harward
stharward@gmail.com

Hi Soren,

2015-10-02 16:31 GMT+02:00 Soren Harward <stharward@gmail.com>:

#1: ./lib/raster/open.c needs to include <errno.h> explicitly. Otherwise
errno is undefined at line 595.

#2: ./general/g.proj/main.c declares the variable "inepsg" only if HAVE_OGR
is defined, but then uses this symbol outside "#ifdef HAVE_OGR" in lines 302
and 322. I'm not exactly sure how to fix this.

first of all, thanks for reporting the issue. The both issues should
be fixed in r66397 in trunk. Please could you test it?

Thanks! Martin

--
Martin Landa
http://geo.fsv.cvut.cz/gwiki/Landa
http://gismentors.cz/mentors/landa

Hi,

2015-10-02 16:54 GMT+02:00 Soren Harward <stharward@gmail.com>:

On Fri, Oct 2, 2015 at 10:46 AM, Martin Landa <landa.martin@gmail.com> wrote:

first of all, thanks for reporting the issue. The both issues should
be fixed in r66397 in trunk. Please could you test it?

That compiles fine. Thanks!

backported to grass70 release branch as r663978. Ma

--
Martin Landa
http://geo.fsv.cvut.cz/gwiki/Landa
http://gismentors.cz/mentors/landa

2015-10-02 19:04 GMT+02:00 Martin Landa <landa.martin@gmail.com>:

backported to grass70 release branch as r663978. Ma

r66398. Ma

--
Martin Landa
http://geo.fsv.cvut.cz/gwiki/Landa
http://gismentors.cz/mentors/landa

Hello,
some of GRASS compilation "options" are not optional at all -
resulting outcome even if will work, will be useless for endusers.
Don't try to push everything into use flags.

I haven't been trying to install GRASS from ebuilds since 6.0.0 days,
still I could test on ~AMD64. Feel free to contact me offlist.
Māris.

2015-10-02 17:31 GMT+03:00 Soren Harward <stharward@gmail.com>:

I'm working on packaging 7.0.1 (well, writing a new ebuild) for Gentoo
Linux. When testing a totally stripped down configuration (ie, everything
disabled that could be), I ran into a couple of compile errors. Here are
things that need to be fixed:

#1: ./lib/raster/open.c needs to include <errno.h> explicitly. Otherwise
errno is undefined at line 595.

#2: ./general/g.proj/main.c declares the variable "inepsg" only if HAVE_OGR
is defined, but then uses this symbol outside "#ifdef HAVE_OGR" in lines 302
and 322. I'm not exactly sure how to fix this.

--
Soren Harward
stharward@gmail.com

--

Soren Harward
stharward@gmail.com

_______________________________________________
grass-dev mailing list
grass-dev@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-dev

On Sat, Oct 3, 2015 at 9:46 AM, Maris Nartiss <maris.gis@gmail.com> wrote:

Hello,
some of GRASS compilation "options" are not optional at all -
resulting outcome even if will work, will be useless for endusers.
Don't try to push everything into use flags.

I haven't been trying to install GRASS from ebuilds since 6.0.0 days,
still I could test on ~AMD64. Feel free to contact me offlist.
Māris.

Please add a summary here as well, like for the other distros:

https://grasswiki.osgeo.org/wiki/Compile_and_Install#Gentoo

Markus

On Sat, Oct 3, 2015 at 3:46 AM, Maris Nartiss <maris.gis@gmail.com> wrote:

Hello,
some of GRASS compilation "options" are not optional at all -
resulting outcome even if will work, will be useless for endusers.
Don't try to push everything into use flags.

That's not really my responsibility, or any developer's responsibility
for that matter. If GRASS exposes compile-time options through the
configure script, then the Gentoo way is to expose those through USE
flags in the ebuild. If the user's install is missing features
because they didn't have their flags set properly, that's on them.

--
Soren Harward