RE: [xml-dev] OOXML: So what *should* happen now?

[robert_weir@us.ibm.com]

Maybe it is considered old fashioned
these days, but I'm accustomed to measuring the sufficiency of a quality
review not by the number of defects found but by the coverage ratio of
the thing being reviewed.  In other words, 3,549 errors from reviewing
20% of a standard may not be so good, but finding 200 errors from reading
100% of it may be very good, if those are the 200 most serious errors.

Have you developed any sense of the
coverage of these comments?  My quick glance seems to indicate that
Part 5 was not reviewed much.  Same for PresentationML.  I certainly
know that my own review never got past the 20% mark for coverage.

-Rob


Rick Jelliffe <rjelliffe@allette.com.au> wrote
on 09/10/2007 08:01:32 AM:

> By the way, the most recent count I have on the number of issues raised
> on DIS 29500 is about 3549. I would imagine there would be many
> duplicates or groupings. So much for "no review"!
> 
> I've quoted recent discussion on this list on my blog
> http://www.oreillynet.com/xml/blog/2007/09/on_error_rates_in_drafts_of_st.html
> 
> Cheers
> Rick Jelliffe