Conflicting definitions of db__driver_describe_table

[hamish]

Hi,

as reported here, ​http://bugs.debian.org/748680

there is a param def'n conflict for db__driver_describe_table() in db/drivers/dbf/describe.c and lib/db/stubs/desc_table.c.

int db__driver_describe_table(dbString * table_name, dbTable ** table)
  vs
int db__driver_describe_table(dbString * name, dbTable * table)

present in all branches, 6.3.0-trunk.

thanks, Hamish

[martinl]

Ticket retargeted after milestone closed

[Bas Couwenberg]

This affects 7.0.2-rc2 too:

db/drivers/odbc/describe.c:int db__driver_describe_table(table_name, table)
db/drivers/odbc/dbdriver.h:int db__driver_describe_table();
db/drivers/postgres/describe.c:int db__driver_describe_table(dbString * table_name, dbTable ** table)
db/drivers/postgres/dbdriver.h:int db__driver_describe_table();
db/drivers/ogr/describe.c:int db__driver_describe_table(dbString * table_name, dbTable ** table)
db/drivers/ogr/dbdriver.h:int db__driver_describe_table();
db/drivers/mysql/describe.c:int db__driver_describe_table(dbString * table_name, dbTable ** table)
db/drivers/mysql/dbdriver.h:int db__driver_describe_table();
db/drivers/sqlite/describe.c:int db__driver_describe_table(dbString * table_name, dbTable ** table)
db/drivers/sqlite/dbdriver.h:int db__driver_describe_table();
db/drivers/dbf/describe.c:int db__driver_describe_table(dbString * table_name, dbTable ** table)
db/drivers/dbf/dbdriver.h:int db__driver_describe_table();
lib/db/stubs/desc_table.c:int db__driver_describe_table(dbString * name, dbTable * table)
lib/db/dbmi_driver/dbstubs.h:extern int db__driver_describe_table();

[neteler]

This is in this case harmless since

lib/db/stubs/desc_table.c

int db__driver_describe_table(dbString * name, dbTable * table)
{
    db_procedure_not_implemented("db_describe_table");
    return DB_FAILED;
}

Magically the DBMI lib works without it. Mysteries of DBMI which we inherited completely undocumented...

Of course it should be updated properly.

[glynn]

Replying to neteler:

The prototype for db__driver_open_select_cursor() is also incorrect.

Of course it should be updated properly.

r66772 adds parameter types to the dbdriver.h and dbstubs.h header files. r66773 changes the lib/stubs/*.c files to include the dbstubs.h header, and fixes the two incorrect prototypes.

Ideally, the individual database drivers should also be changed so that each source file includes the driver's dbdriver.h header.

[neteler]

Travis-CI shows now some issues:

(cd /home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/lib; ln -f -s libgrass_dbmibase.7.1.svn.so /home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/lib/libgrass_dbmibase.so)
if [ "" != "" -a -f "".html ] ; then make html ; fi
make[4]: Leaving directory `/home/travis/build/GRASS-GIS/grass-ci/lib/db/dbmi_base'
make -C dbmi_client || echo /home/travis/build/GRASS-GIS/grass-ci/lib/db/dbmi_client >> /home/travis/build/GRASS-GIS/grass-ci/error.log
make -C stubs || echo /home/travis/build/GRASS-GIS/grass-ci/lib/db/stubs >> /home/travis/build/GRASS-GIS/grass-ci/error.log
make[4]: Entering directory `/home/travis/build/GRASS-GIS/grass-ci/lib/db/stubs'
test -d OBJ.x86_64-pc-linux-gnu || mkdir -p OBJ.x86_64-pc-linux-gnu
make[4]: Entering directory `/home/travis/build/GRASS-GIS/grass-ci/lib/db/dbmi_client'
test -d OBJ.x86_64-pc-linux-gnu || mkdir -p OBJ.x86_64-pc-linux-gnu
gcc  -g -O2  -fPIC  -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include    -DPACKAGE=\""grasslibs"\"   -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -DRELDIR=\"lib/db/stubs\" -o OBJ.x86_64-pc-linux-gnu/add_col.o -c add_col.c
gcc  -g -O2  -fPIC  -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include     -I../dbmi_base -DPACKAGE=\""grasslibs"\"   -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -DRELDIR=\"lib/db/dbmi_client\" -o OBJ.x86_64-pc-linux-gnu/c_add_col.o -c c_add_col.c
add_col.c:2:27: fatal error: grass/dbstubs.h: No such file or directory
compilation terminated.

...

gcc  -g -O2  -fPIC  -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include     -I../dbmi_base -DPACKAGE=\""grasslibs"\"   -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -DRELDIR=\"lib/db/dbmi_client\" -o OBJ.x86_64-pc-linux-gnu/c_list_drivers.o -c c_list_drivers.c
d_error.c: In function ‘db_d_append_error’:
d_error.c:78:11: warning: ignoring return value of ‘fread’, declared with attribute warn_unused_result [-Wunused-result]

...

gcc  -g -O2  -fPIC  -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include     -I../dbmi_base -DPACKAGE=\""grasslibs"\"   -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -I/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/include -DRELDIR=\"lib/db/dbmi_client\" -o OBJ.x86_64-pc-linux-gnu/select.o -c select.c
gcc -shared -o /home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/lib/libgrass_dbmidriver.7.1.svn.so -L/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/lib -L/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/lib -Wl,--export-dynamic -Wl,-rpath-link,/home/travis/build/GRASS-GIS/grass-ci/dist.x86_64-pc-linux-gnu/lib -Wl,-soname,libgrass_dbmidriver.7.1.svn.so OBJ.x86_64-pc-linux-gnu/d_add_col.o OBJ.x86_64-pc-linux-gnu/d_bindupdate.o OBJ.x86_64-pc-linux-gnu/d_close_cur.o OBJ.x86_64-pc-linux-gnu/d_closedb.o OBJ.x86_64-pc-linux-gnu/d_createdb.o OBJ.x86_64-pc-linux-gnu/d_create_idx.o OBJ.x86_64-pc-linux-gnu/d_create_tab.o OBJ.x86_64-pc-linux-gnu/d_delete.o OBJ.x86_64-pc-linux-gnu/d_deletedb.o OBJ.x86_64-pc-linux-gnu/d_desc_table.o OBJ.x86_64-pc-linux-gnu/d_drop_col.o OBJ.x86_64-pc-linux-gnu/d_drop_index.o OBJ.x86_64-pc-linux-gnu/d_drop_tab.o OBJ.x86_64-pc-linux-gnu/d_error.o OBJ.x86_64-pc-linux-gnu/d_execute.o OBJ.x86_64-pc-linux-gnu/d_fetch.o OBJ.x86_64-pc-linux-gnu/d_finddb.o OBJ.x86_64-pc-linux-gnu/d_insert.o OBJ.x86_64-pc-linux-gnu/d_listdb.o OBJ.x86_64-pc-linux-gnu/d_list_idx.o OBJ.x86_64-pc-linux-gnu/d_list_tabs.o OBJ.x86_64-pc-linux-gnu/d_mkdir.o OBJ.x86_64-pc-linux-gnu/d_opendb.o OBJ.x86_64-pc-linux-gnu/d_openinsert.o OBJ.x86_64-pc-linux-gnu/d_openselect.o OBJ.x86_64-pc-linux-gnu/d_openupdate.o OBJ.x86_64-pc-linux-gnu/d_priv.o OBJ.x86_64-pc-linux-gnu/driver.o OBJ.x86_64-pc-linux-gnu/driver_state.o OBJ.x86_64-pc-linux-gnu/d_rows.o OBJ.x86_64-pc-linux-gnu/d_update.o OBJ.x86_64-pc-linux-gnu/d_version.o  -lgrass_dbmibase.7.1.svn -lgrass_dbstubs.7.1.svn -lgrass_gis.7.1.svn -lm 
/usr/bin/ld: cannot find -lgrass_dbstubs.7.1.svn
collect2: ld returned 1 exit status

The first issue will cause the third (and the second is "just" a warning).

Interestingly, changing into lib/db/stubs/ and running make on my local machine for a second time resolved it. Probably only a Makefile issue?

[neteler]

The log file: ​https://s3.amazonaws.com/archive.travis-ci.org/jobs/90080323/log.txt

[hcho]

I'm having the same issue when compiling SVN. I needed to copy lib/db/dbmi_driver/dbstubs.h to dist..../include/grass.

[neteler]

Replying to hcho:

I'm having the same issue when compiling SVN. I needed to copy lib/db/dbmi_driver/dbstubs.h to dist..../include/grass.

for the record, fixed by Glynn in r66817

Question: can r66772 + r66773 + r66817 now be backported?

[martinl]

Replying to neteler:

Replying to hcho:

I'm having the same issue when compiling SVN. I needed to copy lib/db/dbmi_driver/dbstubs.h to dist..../include/grass.

for the record, fixed by Glynn in r66817

Question: can r66772 + r66773 + r66817 now be backported?

Backported to relbr70 in r66884

[martinl]

Backport to relbr64 is not so easy, is someone planning to do it. If not, I suggest to change milestone to 7.0.3 and close the ticket.

[neteler]

Replying to martinl:

Backport to relbr64 is not so easy, is someone planning to do it. If not, I suggest to change milestone to 7.0.3 and close the ticket.

Agreed. Closing for now, feel free to reopen (anyone) if a GRASS GIS 6 patch is provided.