Just testing against (<"Sql","sql","sql_util","module">) should be ok (and much faster).
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
Previous text:
2003-12-20 15:54: Subject: Re: SQL
That makes sense, but would anyone be opposed to me making the constraint more specific? Alternately, is there any opinion about possibly making some changes to allow others to add SQL database support without having to add it to the core? Perhaps Sql.providers.* ?
Bill
On Saturday, December 20, 2003, at 09:25 AM, Johan Sundström (achtung xmas!) @ Pike (-) developers forum wrote:
Most probably so, but we kind of know the ones with [Ss]ql prefixes to be a non-growing static collection, unlike the *_result classes. :-)
/ Johan Sundström (achtung xmas!)
Previous text:
2003-12-20 14:02: Subject: Re: SQL
I think the constraint is there to prevent you from instantating the things in Sql which _aren't_ protocols, which tend to have names starting with "sql":
indices(Sql);
(1) Result: ({ /* 15 elements */ "sybase", "msql", "postgres", "sql_result", "mysql", "sql_util", <-- this one "odbc", "sql", <-- this one "Sql", <-- and this one "rsql", "module", <-- the test seems to miss this one tho "postgres_result", "odbc_result", "oracle", "mysql_result" })
(Also, the names ending with "_result" are excempt for the same reason.)
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
/ Brevbäraren