[GRASS5] Whose oops?

Glynn Clements glynn.clements at virgin.net
Sun Sep 9 23:07:20 EDT 2001

Rich Shepard wrote:

>   Building the weekly snapshot from last Friday (Sept. 7th), all goes well
> until I try to make the postgres links:
> /opt/grass/src.garden/grass.postgresql/d.rast.pg
>   make -f OBJ.i686-pc-linux-gnu/make.rules
> make[1]: Entering directory
> /opt/grass/src.garden/grass.postgresql/d.rast.pg'
> gcc -g -O2  -I/opt/grass/src/include   -c buildInfxQry.c -o
> OBJ.i686-pc-linux-gnu/buildInfxQry.o
> gcc -g -O2  -I/opt/grass/src/include   -c runInfxFile.c -o
> OBJ.i686-pc-linux-gnu/runInfxFile.o
> runInfxFile.c:6: libpq-fe.h: No such file or directory

This is a standard PostgreSQL header. You probably need to re-run
configure with the --with-postgres-includes= switch.

> make[1]: *** [OBJ.i686-pc-linux-gnu/runInfxFile.o] Error 1
> make[1]: Leaving directory /opt/grass/src.garden/grass.postgresql/d.rast.pg'
> make: *** [all] Error 2
>   Is this a user error on my part or a bug in the latest source code?

Technically the former, but it was worked around until a few weeks ago
by configure automatically adding -I switches as it saw fit. 
Unfortunately the workaround meant that certain systems (particularly
those using gcc in conjunction with the vendor's libc headers) could
not be made to build GRASS without manual hacking of the "head" file.

Currently, configure doesn't add any -I switches unless it is told to
by the --with-includes switch or the appropriate --with-*-includes
switch (apart from X_CFLAGS, which is set by AC_PATH_XTRA, typically
by running Imake).

At present, configure only generates warnings for missing libraries
(libpq normally goes in a standard directory; it's only the headers
which are "hidden"). I'll add warnings for missing headers.

Glynn Clements <glynn.clements at virgin.net>

More information about the grass-dev mailing list