drop more than one table in one call

According to http://www.postgresql.org/docs/7.4/interactive/sql-droptable.html, multiple tables can be dropped in one call.
Is this supported by GridSQL ?

No, this is not supported

Is it possible to provide asynchronous drop table operation ?
I.e. drop table request would return immediately without waiting.

Well, it is easier to support multiple names in one "drop table" query.
Feel free to make the change and send us a patch.

First you need to update grammar src/com/edb/gridsql/parser/core/sql_extendb.jj
Then re-generate parser files and modify src/com/edb/gridsql/parser/SqlDropTable.java (update node databases) and src/com/edb/gridsql/metadata/SyncDropTable.java (update metadata)

I plan to add the following to sql_extendb.jj:
void TableNameList(PrintStream prn):
{}
{
(Identifier(prn)|<TEMPDOT_>Identifier(prn)|<PUBLICDOT_>Identifier(prn)|<QPUBLICDOT_>Identifier(prn))
(LOOKAHEAD(2) "," (Identifier(prn)|<TEMPDOT_>Identifier(prn)|<PUBLICDOT_>Identifier(prn)|<QPUBLICDOT_>Identifier(prn)))*
}

I am cvs novice. How can I produce a patch ?
I used this command to check out gridsql/apps:
cvs -z3 -d:pserver:anonymous@gridsql.cvs.sourceforge.net:/cvsroot/gridsql co -P apps

tyumac:gridsql-cvs tyu$ pwd
/Users/tyu/gridsql-cvs
tyumac:gridsql-cvs tyu$ cvs rdiff -c apps
cvs rdiff: No CVSROOT specified! Please use the `-d' option
cvs [rdiff aborted]: or set the CVSROOT environment variable.

I don't see -d option on http://www.manpagez.com/man/1/cvs/

Please advise

Sorry, I have never used cvs directly. My IDE (Eclipse) has built-in cvs support and I use GUI for these tasks.
I guess you shoud either use
cvs -d:pserver:anonymous@gridsql.cvs.sourceforge.net:/cvsroot/gridsql rdiff ...
or define CVSROOT environment variable
export CVSROOT=:pserver:anonymous@gridsql.cvs.sourceforge.net:/cvsroot/gridsql

I plan to declare in sql_extenddb.jj:
void TableNameList(PrintStream prn):
{}
{
(Identifier(prn)|<TEMPDOT_>Identifier(prn)|<PUBLICDOT_>Identifier(prn)|<QPUBLICDOT_>Identifier(prn))
(LOOKAHEAD(2) "," (Identifier(prn)|<TEMPDOT_>Identifier(prn)|<PUBLICDOT_>Identifier(prn)|<QPUBLICDOT_>Identifier(prn)))*
}

This potentially allows temp table(s) to be specified along with other types of tables.
However, in SqlDropTable.execute() I see this:
if (temporary) {
client.deregisterTempTableWithSession(referenceName);
}
If tablename is expanded to a list (array), temporary and referenceName would also need to be expanded.

Please confirm whether this direction is the correct path.

Here is the motivation for my initiative:
The loop in dropTablesInternal() took 23 minutes for the flow run http://build.ciq.com/hudson/view/m2m-perf/job/m2m-staging-grid-incr-regr/8/console:
INFO [2010-08-30 15:08:28] (ExecUtil.java:257) - INFO [2010-08-30 22:08:28] (DatabasePartitioner.java:554) - dropTablesInternal: Dropping table: ROLLUP_1__20100407_tmp
....
INFO [2010-08-30 15:31:22] (ExecUtil.java:257) - INFO [2010-08-30 22:31:22] (DatabasePartitioner.java:554) - dropTablesInternal: Dropping table: ROLLUP_5__20100313_tmp

We want to reduce the amount of time dropping the tables.

You just need to alter dropTable() definition and append ("," TableName(prn))*
No need for lookahead.
To re-generate parser files change to com/edb/gridsql/parser/core amd run make generate.
Then you should update SqlDropTable.java, start from visit() method. Update grammar rule in the comment. Just find generated dropTable class (parameter of visit()) and copy paste comment from there.

Please confirm make command syntax:
make GridSQL DEBUGINFO=-g

This is correct if you making "Debug" version, with debugging symbols info included.
Leave off DEBUGINFO=-g to make "Release" version.