Tom Lane [mailto:tgl@sss.pgh.pa.us] wrote:
>> 2. If the folding mode is chosen through a GUC variable, which
>> is certainly what people would expect, then it turns out that
>> it breaks client libraries/applications *anyway*, because an
>> installation-wide setting could impose itself on a client that
>> hadn't asked for it.
I know that some variables can only be configured at a wide scope, and
not a narrow one. Is there no way to restrict a GUC variable's
configuration scope to session and finer, but force a fixed value at
global scope?
If it is possible to restrict global configuration, that at least
protects the general purpose administrative tools to a significant
degree.
>> And for libraries, that isn't a great solution because then they're
incompatible with applications that wanted another setting.
Good point. Libraries continue to have problems even with session level
configuration if they are to operate in the context of an application
that reconfigures its session case folding for its own purposes.
But, that seems like a problem that is much more likely to affect
developers of new systems rather than general users or administrators of
existing database systems.
If so, it is more of a forward looking problem than a legacy problem in
the sense that the person who encounters it is likely to be in a
position to do something about it. This makes it much less critical to
get every library in the world updated to support all case folding modes
than would be the case for general administrative tools like pgAdmin.
Depending on the nature of the library, a developer would have the
option of using multiple sessions or, perhaps, if it were possible,
modifying the folding configuration when using the library.
Anyhow, as you say, libraries clearly continue to have issues even with
restricted scope on case folding configuration.
And the session level idea really helps nothing unless the global
default session configuration is fixed.
Ian Lewis (www.mstarlabs.com)