On 13 October 2010 13:45, Jan Otto <asche@me.com> wrote:
> we can only prove self-consistency, because there is no algorithm behind
> the scene. the ranges gets applied to publishers depending on how much
> books they publishing over time and probably other criteria.
What about the issue I raised about new ranges coming into use in the
future? That isn't made any worse by your patch, but I'd like to hear
your thoughts on that.
> of course, we can build regression-tests for checkdigits and convert-functions.
> e.g. convert an isbn-10 to isbn-13.
Nothing has changed there. The ISBN-13 checkdigit is the same as
EAN-13 checkdigit (after all, the ISBN-13 is an EAN-13), and the
conversion from ISBN-10 to 13 just involves taking away the bookland
country code (first 3 digits) and changing the checkdigit (last
digit).
Although that's fairly simple, I'd like to hear in more detail how the
regression test will work.
I'd also like to establish just how sensible it is for us to attempt
to hyphenate ISBN-13s. After all, the XML file you linked to has a
timestamp from just two days ago:
<MessageDate>Mon, 11 Oct 2010 15:56:34 GMT</MessageDate>
Maybe it's too late for that though.
--
Regards,
Peter Geoghegan