Re: DBD::XML

On Sat, Aug 20, 2016 at 08:32:09PM -0400, Nigel Horne wrote:
> On 19/8/16 11:51, Tim Bunce wrote:
> > On Fri, Aug 19, 2016 at 10:09:45AM -0400, Nigel Horne wrote:
> > > On 8/19/16 9:56 AM, Tim Bunce wrote:
> > > > On Fri, Aug 19, 2016 at 09:30:32AM -0400, Nigel Horne wrote:
> > > 
> > > <?xml version="1.0" encoding="US-ASCII"?>
> > > <table>
> > >      <row id="1">
> > >          <name>Nigel Horne</name>
> > >          <email>[email protected]</email>
> > >      </row>
> > >      <row id="2">
> > >          <name>A N Other</name>
> > >          <email>[email protected]</email>
> > >      </row>
> > > </table>
> > Does that format ('table', 'row', 'id') correspond with a known XML Schema?
> 
> Nope, that's me creating random test data to poke around.

:)

> > > use DBD::XML;
> > (Ideally users shouldn't need to use the driver module explicitly.)
> 
> I'm assuming after registration that would go away, or am I wrong?

It's not needed now or after registration. (The only time importing a
driver would be useful is if there's a need to import something from
the module, like a bunch of constants.)


> > > my $dbh = DBI->connect('dbi:XML(RaiseError => 1):');
> > > $dbh->func('person', 'XML', "$Bin/../data/person.xml", 'ad_import');  # to be replaced with xml_import once the driver has been registered
> > I presume ad_import comes from DBD::AnyData. Is that 'inspired by',
> > or 'is a fork of', or 'using under the hood'?
> 
> "Pinched from" to get a bootstrap while I'm developing before registration.

:)

> > Are any other XML Schema supported, or supportable?
> I hope so, once I'm ready to create more test data beyond the trivial stuff
> I'm using to get started.
> > Is the XML and/or the parsed data loaded into memory or does each
> > $sth->fetch call pull the next chunk from the XML parser?
> I'm hoping to do chunk by chunk, but that's not done yet.
> > In other words, can it read files larger than the available memory?
> > (Not related to the naming, just curious :)
> I really hope so, but not yet.  I need to walk before I can run :-)

Understood, though I'd caution that it's worth thinking about as these
things may require a different kind of architecture under the hood.

Anyway, back to the topic of naming... I'd suggest something like
DBD::XMLSimpleTable which would have a corresponding prefix of 'xmlst_'.

Sound ok?

Tim.