Zeoslib Portal

Hi Jan, If the cutoff for using INFORMATION_SCHEMA is version 8 and above, note that there's an issue with GetProcedureColumnsFromProcTable. It currently won't work for any procedure where the return value includes charset or collation information. The main reason for my changing the cutoff to 5.7 w...

My earlier assumption that INFORMATION_SCHEMA was added in 5.0 appears to be incorrect. I have the MySQL 3.23, 4.0, 4.1 documentation pdf and it seems that it existed at least as far back as 4.0 (which apparently is when the "show databases" command was added?) Now I'm wondering how far ba...

It's also odd that the Oracle SupportsNonEscapedSearchString has a client version check. I don't think it ever supported an escape string without the escape '\' clause, regardless of the server or client version.

-Mark

It's certainly an interesting issue (and yes you're correct about Oracle.) Really the metadata 'get' functions all serve double duty. They're either 'get' or 'search' depending on if wildcards are there. It's just that detecting the wildcards is iffy due to SQL's underscore issues lol. I suppose the...

It looks like there's no workaround for Oracle (and I assume SQLServer?) I suppose it's not that big of an issue, it would be pretty rare to have multiple tables that would match just based on a single underscore wildcard. I guess it is what it is.

-Mark

Ha! Yes, "something similar" may just be calling that function!

-Mark

Hi Jan, Agreed on using the existence of information_schema to decide to use it. That seems a much better choice. You should see that the metadata changes I pushed don't do the fork check since checking version > 5.0.2 is enough. Note that IsMySQL is also used for other things, although I'm not sure...

Hi Jan, On looking further I see that "internal" metadata calls use AddEscapeCharToWildcards on those arguments, I wasn't aware of that and so was having the issue only in my own testing. I'll look into doing something similar for my "external" metadata calls that don't need wild...

Hi All! Since the various "gettables" , "getprocedures", "getcolumns" and such metadata calls allow passing in a pattern, and the ConstructNameCondition call is often used in creating the data dictionary queries, we run into an issue where any table with an underscore c...

Onions! Onions have layers! Yes, just one layer for Oracle! I'm not sure what I was thinking there. I don't think of Firebird database as a layer because I don't use them in dot notation when referring to objects (the dot notation the database uses for referring to a table is my (possibly goofy) &qu...

Hi Michael, If I read that document correctly they'll be adding a "catalog" layer above the existing MySQL "database" layer (which they've already done in the information_schema.) So you'll be able to refer to a table using catalog.database.tablename whereas now it's database.tab...

Agreed on all points! I have an update ready to go for MySQLMetaData which has several fixes and improvements, but I'm just wondering about what to return in the catalog and schema fields of tables. The current one returns database in CATALOG and nothing in Schema. It's inconsistent with the other m...

The catalog vs schema thing isn't too bad I think. The real nightmare is case sensitivity in MySQL since it depends on the OS filename case sensitivity and the lower_case_table_names variable setting. It's just brutal.

-Mark

The terms are a bit different depending on the database you use, however the Zeos metadata layer works with the terms (as parameters and object fields) "Catalog" and "Schema", so it's just a matter of deciding how these map to the underlying data dictionary. I usually think of SC...

In the Metadata calls for tables and procedures there's an inconsistency in how catalog and schema are handled: Currently for Tables, Columns, TablePrivs, ColumnPrivs, Pks, Indexes: TABLE_CAT = Database TABLE_SCHEMA = Empty Currently for Procedures, ProcedureColumns: PROCEDURE_CAT = 'def' (the value...