On Nov 29, 2007 6:01 PM, Ben Barrett <[EMAIL PROTECTED]> wrote:

> Thanks Marbux,
> I was [partially] surprised that they said Excel was the biggie, I heard
> from a CCC conference, albeit a while back, that Vista had all their old
> known issues, at least at the outset, that they had identified about 90% of
> the most-buggy DLLs were identical to XP on the inside :)
>

I wonder whether that part is accurate. It used to be that around 50 per
cent of the known malware out there was embedded Word scripts. And Word has
a much larger user base than Excel.

On the persistent of bugs in Windows DLLs, it's much the same story with MS
Office. Its major apps were designed back in the days when Bill Gates used
to brag about not fixing bugs. Now they're in the spot 15 years later where
it's too late; gobs of spaghetti code on top of the original bugs. There's a
study I did in 2000 of all the published bugs in the Word footnote and
endnote features here, <http://www.llrx.com/features/word.htm>. If you skim
the tables, you'll see a lot of bugs, even data loss bugs, that have
persisted for many versions. And you can infer from the version numbers for
some of the persistent bugs that Word still has its original 16-bit page
layout engine. (Some of those versions ran on 80086 and 80088 processors.)

I haven't studied bugs in Excel and Powerpoint so closely, but I strongly
suspect it's much the same  story there too. E.g., at ISO, Microsoft is
lobbying hard to be allowed to keep the Excel leap year bug in the OOXML
specification. I doubt if they'd be defending the bug so vigorously if they
could repair it other than by using pre- and post-processors, which would
give them a big performance hit in loading large spreadsheets.

Best regards,

Marbux
_______________________________________________
EUGLUG mailing list
[email protected]
http://www.euglug.org/mailman/listinfo/euglug

Reply via email to