I was doing testing on an app and wanted to make a table name a variable
so I can change it via a config file (instead of re-compile/relink).
like so: EXEC SQL INSERT INTO :tname (host, point, type, cooked) VALUES (:host, :name, :type, :cooked);
This fails ecpg preprocess with "ERROR: syntax error at or near ":tname" whereas if the table
name is specified statically it works fine, like: EXEC SQL INSERT INTO tname (host, point, type, cooked) VALUES (:host,
:name,:type, :cooked);
When I tried hacking the resulting C code (as a test exercise) like:
{ ECPGdo(__LINE__, 0, 1, NULL, "insert into ? ( host , point , type , cooked ) values ( ? , ? , ? , ? )",
ECPGt_char,&(tname),(long)0,(long)1,1*sizeof(char),ECPGt_char,&(host),(long)0,(long)1,1*sizeof(char),
ECPGt_NO_INDICATOR,NULL , 0L, 0L, 0L, ECPGt_char,&(name),(long)0,(long)1,1*sizeof(char), ECPGt_NO_INDICATOR, NULL , 0L,
0L,0L, ECPGt_char,&(type),(long)1,(long)1,1*sizeof(char), ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L,
ECPGt_char,&(cooked),(long)1,(long)1,1*sizeof(char),ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L,
ECPGt_char,(starttime),(long)30,(long)1,30*sizeof(char),ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L,
ECPGt_char,(interval),(long)30,(long)1,30*sizeof(char),ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L,
it got all funny and didn't work very well (well ok it didn't work at all): pg.pgc:129 SQL Insert failed: Invalid
statementname (null) in line 124.
So, I was wondering am I just missing some obvious way to do this, or
is this not supported at all?
Thanks
--
>-=-=-=-=-=-=-<>-=-=-=-=-=-<>-=-=-=-=-=-<>-=-=-=-=-=-<>-=-=-=-=-=-=-<
Ryan Mooney ryan@pcslink.com
<-=-=-=-=-=-=-><-=-=-=-=-=-><-=-=-=-=-=-><-=-=-=-=-=-><-=-=-=-=-=-=->
