ZDNet UK

CNET NETWORKS UK BUSINESS SITES:
atlarge.com
BNET UK
silicon.com
SmartPlanet.com
ZDNet.co.uk

ZDNet.co.uk - Winner of Best Business Website 2007

You are here: ZDNet.co.uk > News > Software

Office applications Toolkit

Story: Google: OOXML 'insufficient and unnecessary'

Posted by: Anonymous123 (Wednesday 27 February 2008, 8:39 PM)

Thank you for an intelligent response.

Thank you, Albert, for an intelligent response, and for taking the time to do so. So let's discuss this. I'll try to keep it short and to the point.

Tag name length is a matter of debate to be sure. XML by nature, however, is intended to allow for text editing. Too short names introduce possible confusion over intent, and possible name rangling in the future over additions. Having a tag like "CannonicalCellContents" is ridiculously long, but having a tag like "C" is ambiguously short.

That's why, in my opinion, discussing XML efficiency in terms of tag length is a red herring. XML is what it is, and truely there is a fadishness about it. That is why, again in my opinion, if one is concerned about tag creation/parsing performance, one should go with a binary format. It's not like one cannot create portable binary. Lua does it. I've worked with the byte code and it handles nuxi issues just fine. I'm sure Java does as well. I'd be for BODF (Binary ODF). The design also isn't particularly complicated. (Again, look at lua.)

As for formula tracking, I understand your point, although I'd have to see the tracking mechanism to be able to discuss this intelligently. On the other hand, I'm assuming that with OOXML you need to read in the whole document before doing any processing since tags in general can occur in any order, so what you are really describing is the internal representation of the document, and that doesn't have to have much of any relationship to the XML at all. The real issue here is whether there is any unnecessary duplication of tag values in the XML for formulas. If there is, then a reference mechanism should have been put into ODF, and that is certainly something that could be presented to the committee.

Lastly, as for relashionship information, again I'd have to see the scheme, but again, if it isn't outlandish/Microsoftish that could also in some way end up in the standard. You said that with ODF it is spread all over the document. Depending on how this is used, this implies that some XML structure rearrangement should have been done. And again, that is a reasonable thing to present to the committee. In the end, structured documents are databases, and are subject to the same indexing issues as databases.

To conclude, however, both the formula issue and relationship information issue you present are abstract issues, and a design should be approached from that standpoint. I suspect the problem here is that it wasn't.

View complete profile

Private message disabled

View all posts

Anonymous123
n/a
Member since: February 2008

Site Activity Rating:

Full Talkback thread

Story: Google: OOXML 'insufficient and unnecessary'

Back to: Google: OOXML 'insufficient and unnecessary'

More In Office applications

Featured Talkback

Why do so many (virtually all) software packages think that they are so important that they have to be started automatically every time the computer boots? What is the largest number of "speed access", "update check", "camera download" and whatever other background programs you have ever seen running? Of those, how many did you really need?