Subject | Re: Case-insensitive selects |
---|---|
Author | peter_jacobi.rm |
Post date | 2004-03-07T13:16:20Z |
Hi Thomas,
--- In firebird-support@yahoogroups.com, "Thomas Leggett" :
http://www.jodelpeter.de/i18n/fbarch/index.htm
There is a minimal working example fbintl2.dll with source.
I find it less obscure than the older examples provided by Dave,
but beauty is in the eye of the observer, isn't it?
At http://www.brookstonesystems.com/CollateKit.zip
you find Dave's essential (read: the only one) information about IB/FB
I18n.
Also you should download the source tree of FB1.5 (or FB2.0). The
sources for the system collations (fbintl.dll) are in the src/intl
subdirectory.
If compiling your own distribution of FB1.5 is an option for you,
patching the source tree may be the solution. This will have two
advantages:
- you make your new collations known from startup and no manipulation
of the system tables are necessary
- you can use the existing subroutines and essentially only need to
clone one source file, changing the data table defining the collation,
and you're done.
Starting with FB1.5, I assume, the "user" collations in fbintl2.dll
cannot "call back" into the engine, so all code there must be self
contained. I'm under the impression, that's the point that broke
Dave's no-case collations.
Regards,
Peter Jacobi
--- In firebird-support@yahoogroups.com, "Thomas Leggett" :
> Could you please direct me to the source-code example so we can try toStart reading at my page:
> figure something out?
http://www.jodelpeter.de/i18n/fbarch/index.htm
There is a minimal working example fbintl2.dll with source.
I find it less obscure than the older examples provided by Dave,
but beauty is in the eye of the observer, isn't it?
At http://www.brookstonesystems.com/CollateKit.zip
you find Dave's essential (read: the only one) information about IB/FB
I18n.
Also you should download the source tree of FB1.5 (or FB2.0). The
sources for the system collations (fbintl.dll) are in the src/intl
subdirectory.
If compiling your own distribution of FB1.5 is an option for you,
patching the source tree may be the solution. This will have two
advantages:
- you make your new collations known from startup and no manipulation
of the system tables are necessary
- you can use the existing subroutines and essentially only need to
clone one source file, changing the data table defining the collation,
and you're done.
Starting with FB1.5, I assume, the "user" collations in fbintl2.dll
cannot "call back" into the engine, so all code there must be self
contained. I'm under the impression, that's the point that broke
Dave's no-case collations.
Regards,
Peter Jacobi