On Sat, Mar 16, 2019 at 3:15 PM Tom Lane <tgl@xxxxxxxxxxxxx> wrote:
Steve Haresnape <s.haresnape@xxxxxxxxxxxxxxxxxxxxxxx> writes:
> As I said, I don't want to quote my identifiers. I know what that does. I
> want to specify them in a certain way, see them in that same way, but refer
> to them in any old way.
> You can call it normalize or fold or whatever. It's a bad design choice,
> and not even a completely compliant choice.
> Is a cure contemplated? I know it's not just me that dislikes this.
No.
There have been previous discussions of allowing variant case-folding
rules, and the conclusion has always been that it would break so much
stuff as to be entirely not worth the trouble.
The big problem with making significant semantics changes like this
be optional is that authors of general-purpose tools then have to be
prepared to cope with all the possibilities. That's a pretty enormous
cost to load onto other people. If it *only* affected the core code,
maybe you could find somebody to do the work and call it done, but
actually the implications would reverberate across the entire Postgres
ecosystem. That's a tough call to make for a change that can't even
be painted as meeting a widely-favored goal like better SQL spec
compliance.
Yeah. I remember the annoyances caused by the removal of the implicit casts in 8.3 and I think this would be orders of magnitude worse. The cast changes were clearly worth the (significant) pain. I seriously doubt changing case folding would.
Now, in the spirit of full disclosure, I should say that the only form
of this idea that people have really spent significant effort looking
at is exactly the fully-SQL-spec-compliant case-folding rule, ie just
like Postgres normally does it except unquoted identifiers fold to
all-upper-case not all-lower. Perhaps there's some reason why what
you want would be less painful than that turns out to be ... but I'm
not seeing such a reason offhand. In fact I suspect your preference
is actually worse, it'd require behavior changes in more places.
As an example, I believe your request would require case-insensitive
uniqueness enforcement in the system catalogs' unique indexes on names.
You have no idea how large a can of worms that opens (but I'll just
mention that "which characters are letters" doesn't even have a well
defined universal answer).
+1
regards, tom lane
Best Wishes,
Chris Travers
Efficito: Hosted Accounting and ERP. Robust and Flexible. No vendor lock-in.
