> - why do we need to pop the entire stack in the destructor of the loader > (shouldn't it be always empty)?
It won't be empty if an exception is thrown. > - can the huge number of catch clauses in the loader be reduced by only > catching a base class without loss of semantics? No. Actually, I was going to open another bug after this was merged (since the bug existed before my changes). The old (and still current) code catches the exception and throws a new exception passing the existing what() message -- that's wrong since the what() message from the json::* exceptions is only in English (which is fine since json::* exceptions should never be seen by the user anyway). There really need to be diagnostics entries in the XML file for all the error messages. Also, why are all JSON parser errors squished into the single JNDY0021 error code? Error codes are free. Why not have one error code per parser error (just as I did with my JSON module)? > - changelog entry is missing (for the new feature and the two fixed bugs) What bugs? > I added Ghislain as a reviewer. He needs to make changes to the JSONiq spec > and, > hence, at least needs to agree on the name of the option. I was going to add him as a reviewer myself when I was "done" (which I'm not quite). I was trying to spare him the useless spam e-mail from launchpad until I was "done." So much for me trying to be nice. -- https://code.launchpad.net/~paul-lucas/zorba/feature-json_strip_array/+merge/136826 Your team Zorba Coders is subscribed to branch lp:zorba. -- Mailing list: https://launchpad.net/~zorba-coders Post to : firstname.lastname@example.org Unsubscribe : https://launchpad.net/~zorba-coders More help : https://help.launchpad.net/ListHelp