discuss-gnustep
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: getting started with gdl2


From: David Ayers
Subject: Re: getting started with gdl2
Date: Tue, 24 Feb 2004 16:10:30 +0100
User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6) Gecko/20040113

Richard Frith-Macdonald wrote:


On 23 Feb 2004, at 23:14, David Ayers wrote:

Indeed, Matt also noticed that it works when they're quoted. I'll hope to fix this via sqlStringForSchemaObjectName: / useQuotedExternalNames & Co. throughout GDL2 and the adaptor code. (I'm also considering defaulting the Postgres adaptor too using the quoting /unless/ it is explicitly turned off via the EOAdaptorQuotesExternalNames default, in contrast to the documentation of the abstract class.)


I think I read that this is a standard feature of SQL ... that only quoted names are case sensitive.

I've just attempted to parse SQL99 (2) but only with limited success. (In particular 5.2 21)-28) seems relevant, but I'm not sure yet.) There is discussion about case sensitivity, but I think I'll need to spend some more time with before I really can grok what it's really trying to say.

Yet if this is a standard feature then I'll keep 'NO' as the default for EOAdaptorQuotesExternalNames.

IIRC, postgres deviates from the standard in that unquoted names are made lowercase while they should strictly be made uppercase ... but I think lowercase is more readable, so I'm not going to complain.

In any case, I'd like use the EOAdaptorQuotesExternalNames default to determine whether to use quotes in EOSQLSchemaCreation code, even though EOF doesn't seem to do this. (Note that sqlStringForSchemaObjectName: and externalQuoteCharacter are instance methods, whereas set/useQuotedExternalNames: and all the schema creation methods are class methods. As a consequence EOModeler also doesn't seem to generate the quotes when the default is set.)

I guess we could create temporary bogus expressions in the class methods to generate that part of the SQL to cause less grief for our users. If anyone has a better idea, please let me know.

Cheers,
David

PS: Does anyone know of an online version of SQL99 we could use as reference?





reply via email to

[Prev in Thread] Current Thread [Next in Thread]