If you don't know that it's _true_ then, yes, please remove it.
> You added a section about SQL Identifiers that mentions
> quotes:
>
> But I find this surprising, and I doubt that it's true:
> http://git.gnome.org/cgit/libgda/tree/doc/C/libgda-4.0-docs.sgml#n465
> "
> Many people consider that using the quoted syntax should be
> avoided if
> possible because:
> - once an SQL identifier has been defined using the quoted
> syntax, it
> must always be used quoted, which is error prone especially if
> external
> tools are used on the database.
> "
>
> From my own experience and from what I've gathered from other people
> and from the misc docs and books, this is a kind of personnal
> recommendation I'd do. I added it in the doc to explain that Libgda
> could not force people to use quotes all the time. I can remove it if
> you think it's out of topic or is out of Libgda's scope.
If it's not true then I am concerned that it's the basis of your API
choices.
Oh, I suggested in bugzilla that you keep these changes out of stable
> 2.
> The latest documentation is not online yet because it is not
> yet in a
> tarball, so I can't link to it, but I've built it locally.
>
> I still can't see anything in the
> gda_connection_update_meta_store() (or
> anything linked from there) about whether or not we should use
> gda_sql_identifier_add_quotes() for the "table_name" value in
> GdaMataContext::column_values.
>
> The "Update the meta data about a table" section doesn't use
> gda_sql_identifier_add_quotes(), though we now seem to need to
> use it in
> libgdamm. Note that that section mentions a "Get information"
> section,
> but there is no link.
>
> 3.
> In fact, I don't see any link from GdaMetaContext to a
> description of
> possible values for the GdaMetaContext field structs, though
> the
> gda_connection_update_meta_store() documentation mentios
> "_tables" and
> "_columns" as possible values for GdaMetaContext::table_name.
>
> The "database structure" section provides some hints, but the
> documentation should explain how that maps to GdaMetaContext."
>
> Sorry I've been so long to answer, but I've been busy fixing the SQL
> identifiers mess (there are so many differences between databases in
> details...)
> Anyway, I've just pushed in git LIBGDA_4.0 branch,
4.0, because it all seems so fragile. But OK, if it works it will be
useful there.
This is all in master (4.1) too, right?
Thanks. I'll take a proper look when the documentation is at
> some fixes in the gda_sql_identifier_quote() function which you can
> use in Glom, along with a test case (which I'll soon add to the NR
> tests directory), attached (for the PostgreSQL and MySQL you need to
> create an empty "testcase" database).
>
> Basically, you have an SQL Id (presumably entered by the user), and
> use gda_sql_identifier_quote(id,...,FALSE,...) to have a valid SQL
> identifier to create a database object (for example a table), and use
> again gda_sql_identifier_quote(id,...,TRUE,...) with
> gda_connection_update_meta_store() or other GdaMetaStore's related
> functions. See the test case in the attachment.
>
> I'll apply the modifications to the master branch and fix all the
> documentation which needs to be fixed ASAP.
library.gnome.org (when a tarball has been released).