Dimitry Sibiryakov

[SQL Fiddle](http://sqlfiddle.com/#!4/65ae9e/1) insists that it works ONLY in MySQL and no other DBMS.

Also as a public header it should use either standard or publicly defined Firebird types instead of internally defined types.

> I'm not sure I understand the problem. Padding in the middle of structures to ensure offsets of members is fine. Padding at the end of structures to ensure sizes...

> But how is it possible then to guarantee the size of a struct if the end part of the > struct can vary in size? For such structures this...

Because Alex did it this way. See my PS above.

I would use it like this: ``` void ODBCAttachment::detach(Firebird::CheckStatusWrapper* status) { internalDetach(status); #if FB_API_VER > 40 release(); #endif } ``` ``` #if FB_API_VER > 40 void ODBCAttachment::deprecatedDetach(Firebird::CheckStatusWrapper* status) { internalDetach(status);...

It is impossible because returning types from ES cannot be determined at compile time.

I would suggest to derive the type from the second argument if the first is NULL. Otherwise it will be impossible to gather a binary BLOB without creating of an...

Except it will append data to the bin_blob, no.

And this returns us to the original discussion when I complained about inconsistent unpredictable behavior regarding modification of the first argument.