Thread: BUG #4549: ecpg produces code that don't compile
The following bug has been logged online: Bug reference: 4549 Logged by: Clemens Fischer Email address: cfi@mbs-software.de PostgreSQL version: 8.3.5 Operating system: QNX 6.3 Description: ecpg produces code that don't compile Details: Hello List, Don't know whether anyone here can help but... We have some code that has compiled and ran just fine from postgresql 7.4.x thru 8.2.9. It uses embedded sql. With the new port - 8.3.5 - this code will no longer compile. Below is a standalone code, which shows you the details. Any help or ideas would be appreciated. -- snip ecpgtest-multidim.pgc begin -- #include <stdio.h> #include <stdlib.h> EXEC SQL include sqlca; EXEC SQL WHENEVER SQLERROR sqlprint; EXEC SQL WHENEVER SQLWARNING sqlprint; EXEC SQL WHENEVER NOT FOUND continue; int main( void ) { EXEC SQL begin declare section; char multidim_char[4][50]; // works VARCHAR multidim_varchar[4][50]; // failed EXEC SQL end declare section; ECPGdebug(1, stderr); EXEC SQL connect to postgres user postgres using postgres; EXEC SQL DROP TABLE IF EXISTS ecpgtest; EXEC SQL CREATE TABLE ecpgtest (no INTEGER PRIMARY KEY, name VARCHAR(30) NOT NULL); EXEC SQL INSERT INTO ecpgtest VALUES (1, 'first'); EXEC SQL INSERT INTO ecpgtest VALUES (2, 'second'); EXEC SQL INSERT INTO ecpgtest VALUES (3, 'third'); // test 1 EXEC SQL SELECT name INTO :multidim_char[0] FROM ecpgtest WHERE no = 1; // test 2 EXEC SQL SELECT name INTO :multidim_varchar[0] FROM ecpgtest WHERE no = 1; EXEC SQL DISCONNECT ALL; ECPGdebug(0, stderr); return( EXIT_SUCCESS ); } -- snip ecpgtest-multidim.pgc end -- Using ecpg and gcc generates the following output: ecpg -t -o ecpgtest-multidim.c ecpgtest-multidim.pgc cc -c -I/usr/local/pgsql/include ecpgtest-multidim.c ecpgtest-multidim.pgc: In function `main': ecpgtest-multidim.pgc:30: parse error before `_13' cc: /usr/qnx630/host/qnx6/x86/usr/lib/gcc-lib/ntox86/2.95.3/cc1 error 33 make: *** [ecpgtest-multidim.o] Error 1 -- snip ecpgtest-multidim.c code fragment 1 begin -- int main( void ) { /* exec sql begin declare section */ // works // failed #line 12 "ecpgtest-multidim.pgc" char multidim_char [ 4 ] [ 50 ] ; #line 13 "ecpgtest-multidim.pgc" struct varchar_multidim_varchar_13 { int len; char arr[ 50 ]; } multidim_va rchar [4] ; /* exec sql end declare section */ #line 14 "ecpgtest-multidim.pgc" ECPGdebug(1, stderr); -- snip ecpgtest-multidim.c code fragment 1 end -- You see in "#line 13" the definition of the varchar. -- snip ecpgtest-multidim.c code fragment 2 begin -- // test 1 { ECPGdo(__LINE__, 0, 1, NULL, 0, ECPGst_normal, "select name from ecpgtest where no = 1 ", ECPGt_EOIT, ECPGt_char,(multidim_char[0]),(long)50,(long)1,(50)*sizeof(char), ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L, ECPGt_EORT); #line 27 "ecpgtest-multidim.pgc" if (sqlca.sqlwarn[0] == 'W') sqlprint(); #line 27 "ecpgtest-multidim.pgc" if (sqlca.sqlcode < 0) sqlprint();} #line 27 "ecpgtest-multidim.pgc" // test 2 { ECPGdo(__LINE__, 0, 1, NULL, 0, ECPGst_normal, "select name from ecpgtest where no = 1 ", ECPGt_EOIT, ECPGt_varchar,&(multidim_varchar[0]),(long)50,(long)1,sizeof(struct varchar_multidim_varchar[0]_13), ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L, ECPGt_EORT); #line 29 "ecpgtest-multidim.pgc" -- snip ecpgtest-multidim.c fragment 2 end -- In the second fragment you see the wrong sizeof-call: varchar_multidim_varchar[0]_13 one solution is: varchar_multidim_varchar_13[0] Can you fix this, please? Thank you Clemens Fischer
On Tue, Nov 25, 2008 at 10:44:32AM +0000, Clemens Fischer wrote: > Using ecpg and gcc generates the following output: > ecpg -t -o ecpgtest-multidim.c ecpgtest-multidim.pgc > cc -c -I/usr/local/pgsql/include ecpgtest-multidim.c > ecpgtest-multidim.pgc: In function `main': > ecpgtest-multidim.pgc:30: parse error before `_13' > cc: /usr/qnx630/host/qnx6/x86/usr/lib/gcc-lib/ntox86/2.95.3/cc1 error 33 > make: *** [ecpgtest-multidim.o] Error 1 Thanks for reporting. Fixed in CVS HEAD and 8.3. Patch is attached, it was just a small oversight that happened when the new varchar naming was introduced. Michael -- Michael Meskes Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org) Michael at BorussiaFan dot De, Meskes at (Debian|Postgresql) dot Org ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: meskes@jabber.org Go VfL Borussia! Go SF 49ers! Use Debian GNU/Linux! Use PostgreSQL!
Attachment
Hallo Michael, thanks for the patch, but ... compiling ecpg fails: gcc -O2 -Wall -Wmissing-prototypes -Wpointer-arith -Winline -fno-strict-aliasing -Wno-error -I../include -I../../../../src/interfaces/ecpg/include -I. -DMAJOR_VERSION=4 -DMINOR_VERSION=4 -DPATCHLEVEL=1 -I../../../../src/include -D_GNU_SOURCE -I/opt/include -c -o type.o type.c type.c: In function `ECPGdump_a_simple': type.c:356: warning: implicit declaration of function `strchrnul' type.c:356: invalid type argument of `unary *' make: *** [type.o] Error 1 Is it possible to expand 'strchrnul' to if(var_name != NULL) *(strchr( ... or something like that? Clemens Am Mittwoch, den 26.11.2008, 14:20 +0100 schrieb Michael Meskes: > On Tue, Nov 25, 2008 at 10:44:32AM +0000, Clemens Fischer wrote: > > Using ecpg and gcc generates the following output: > > ecpg -t -o ecpgtest-multidim.c ecpgtest-multidim.pgc > > cc -c -I/usr/local/pgsql/include ecpgtest-multidim.c > > ecpgtest-multidim.pgc: In function `main': > > ecpgtest-multidim.pgc:30: parse error before `_13' > > cc: /usr/qnx630/host/qnx6/x86/usr/lib/gcc-lib/ntox86/2.95.3/cc1 error 33 > > make: *** [ecpgtest-multidim.o] Error 1 > > Thanks for reporting. Fixed in CVS HEAD and 8.3. > > Patch is attached, it was just a small oversight that happened when the new > varchar naming was introduced. > > Michael
"Clemens A. Fischer" <clemens.fischer@mbs-software.de> writes: > Hallo Michael, > thanks for the patch, but ... > compiling ecpg fails: > type.c: In function `ECPGdump_a_simple': > type.c:356: warning: implicit declaration of function `strchrnul' > type.c:356: invalid type argument of `unary *' Whatever strchrnul is, it isn't very portable --- the buildfarm is breaking out in red too. Please change that. regards, tom lane
Tom Lane wrote: > Whatever strchrnul is, it isn't very portable --- the buildfarm is > breaking out in red too. Please change that. >=20=20=20 sigh, more gnuish code.... =97 Function: char * *strchrnul* (const char *string, int c) |strchrnul| is the same as |strchr| except that if it does not find the character, it returns a pointer to string's terminating null character rather than a null pointer. This function is a GNU extension.
On Wed, Nov 26, 2008 at 06:44:11AM -0800, John R Pierce wrote: > Tom Lane wrote: >> Whatever strchrnul is, it isn't very portable --- the buildfarm is >> breaking out in red too. Please change that. > > sigh, more gnuish code.... Fixed. Sorry, didn't notice that it still is a GNU extension as GNU has for quite a long time. Michael -- Michael Meskes Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org) Michael at BorussiaFan dot De, Meskes at (Debian|Postgresql) dot Org ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: meskes@jabber.org Go VfL Borussia! Go SF 49ers! Use Debian GNU/Linux! Use PostgreSQL!