RE: [PEAR-DEV] XML_Query2XML: ISO/IEC 9075-14:2005 mapping

Hi Jesus, > Jesus M. Castagnetto wrote: > > Hi Lukas > > (Sorry I could not answer your message before, I was > preparing some classes and talks) > > --- Lukas Feiler <lukas@empoweredmail.com> wrote: > > > Hi, > > > > as Justin and Jesus suggested, I implemented ISO/IEC > 9075-14:2005 for > > mapping SQL identifies to XML names within XML_Query2XML (voting in > > progress). > > > > I have 2 questions regarding the implementation: > > > > 1) What do you think about (not) making > XML_Query2XML_ISO9075Mapper the > > default? > > ISO 9075-mapping is quite complex as it also requires > Unicode support. > > Therefore my full implementation has 356 lines and on a > PIII, 2.2 GHz it > > does only about 1755 mappings per second. Considering this > overhead, I'm not > > sure if XML_Query2XML_ISO9075Mapper should be the default > mapper. I guess a > > lot of people only use [a-Z_0-9]* for SQL identifiers and > don't need any > > special mapping at all. > > Agree on this one. The extended mapping will be useful in the cases of > accessing some legacy or foreign language named identifiers, > thus it should not > be the default but an option. I've dealt with some old dbs > that included spaces > and other random stuff in the names, just because of some > internal standard > that mandated a fixed number of chars, so the db admins got > inventive in their > naming schemes :-) Not using ISO9075 mapping as the default also has the advantage that the dependency on I18N_UnicodeString is only optional. Users that don't do ISO9075 mapping won't even have to install I18N_UnicodeString (which is currently only beta). > > > 2) Why do you think that Section 9.2 of ISO/IEC > 9075-14:2005 is relevant? > > XML_Query2XML only deals with column names but never with > multiple SQL > > identifiers that would have to be mapped to a single XML name. > > I would not implement that just yet. I am not sure what could > be the exact > applicability of that in most day to day work. Seems to me > that the usual > mapping of database names, tables and fields in a hierachical > manner in XML > should suffice for most purposes, no need to have a multipart > fo!!.ba++r.bo1#ng > mapping to a single element. OK, I agree > > > > > The source of ISO9075Mapper.php > (XML_Query2XML_ISO9075Mapper) can be found > > at: > > http://query2xml.sourceforge.net/ISO9075Mapper.php.html > > > > Unit tests can be found at: > > > > http://query2xml.sourceforge.net/ISO9075MapperTest.php.html > > > > Note: currently XML_Query2XML_ISO9075Mapper has its own methods for > > converting characters from UTF-8 to Unicode (and vice > versa); as suggested > > by Justin, I placed a feature request for the PEAR package > > I18N_UnicodeString. This package actually contains the required > > functionality but does not currently provide it via public > static methods. > > I see that your request has been attended so you can now use > the methods in the > aforementioned package instead of the internally defined ones > in your code. > > I am impressed with having implemented this, and that you had > the stamina to > waddle through that standard (which is not light reading at all). > > Cheers and keep the good work. Thanks for making me implement ISO 9075 :) It's a great feature! As soon as the PEPr voting process is over I'll make a new release that includes all changes suggested during the voting process. I already implemented and documented everything but I'd like to wait if anything else turns up until tomorrow 08:00 UTC. I just updated http://query2xml.sourceforge.net/ISO9075Mapper.php.html so you can have a look at the new implementation using I18N_UnicodeString. Lukas