David Overton's Blog and Discussion Site
This site is my way to share my views and general business and IT information with you about Microsoft, IT solutions for ISVs, technologists and businesses, large and small.  
Fix for Excel 2007 and Excel Services calculation errors

I know that a few people had the pleasure of throwing a few pies at Microsoft when this was discovered, but a fix is now available.

Description of the Excel 2007 hotfix package: October 9, 2007

This hotfix package fixes the following issue that was not previously documented in a Microsoft Knowledge Base article:
  • When you perform a calculation in Excel 2007, the following behavior occurs:
  • The result of the calculation is a number from 65534.99999999995 to 65535. The calculation is performed correctly. However, the result is incorrectly shown as 100000.
  • The result of the calculation is a number from 65535.99999999995 to 65536. The calculation is performed correctly. However, the result is incorrectly shown as 100001.

 

To resolve this problem, apply this hotfix. After you apply the hotfix, recalculate the spreadsheets that are in manual calculation mode. When you do this, the values in the spreadsheet will be updated to the correct values.

Excel 2007

Download the Excel 2007 hotfix package now. (http://download.microsoft.com/download/6/1/3/61343075-aa12-4152-a761-fccc16d6cef4/office-kb943075-fullfile-x86-glb.exe)

Excel Services

Download the 32-bit Excel Services 2007 hotfix package now. (http://download.microsoft.com/download/c/d/c/cdcccd84-86cd-4199-b01c-1df2dac66534/office-kb943076-fullfile-x86-glb.exe)
Download the 64-bit Excel Services 2007 hotfix package now. (http://download.microsoft.com/download/c/d/c/cdcccd84-86cd-4199-b01c-1df2dac66534/office-kb943076-fullfile-x64-glb.exe)

 

ttfn

 

David

Technorati Tags: , ,

Posted Wed, Oct 10 2007 3:24 PM by David Overton
Filed under: ,

Comments

AdamV wrote re: Fix for Excel 2007 and Excel Services calculation errors
on Wed, Oct 10 2007 11:06 PM

I must admit to being disappointed in the lack of coverage this bug received from MS. Nothing on the Excel 'marketing' pages, not a mention under "new features".

The only place it seemed to get out was via a single MSDN blog, and several independent sources. Even our esteemed Mr. Strange (blogs.msdn.com/.../default.aspx) did not see fit to point it out to people in case they wanted to check whether this might impact on their rollout or their business modelling spreadsheets.

I accept that the chances of this making a real difference are minuscule, but they could have been seriously important for anyone affected (a bit like playing one-in-a-million Russian roulette - if you get the bullet, you are just as dead as if it was 1 in 6).

When Ford (or whoever) screw up something on a new car they have to send out recall notices, or offers to fix the problem at no cost to the owner. Most people would be blissfully unaware of some of the minor issues that trigger such notices. This cost serious money to manage, as well as having a slight impact on their reputation. But a lot less than if things were left to be found out and publicised by others.

I have a client who was thinking very hard about delaying their 1st December start date for rolling out Office 2007 because of this bug - amongst several thousand users they could not be sure this would not occur in one of its forms and remain undetected. Since they are piping data from Excel into an ERP system, fixing the mistake the following month would not work - last month's figures are already posted. You still have to find the individual line errors and correct them.

Since the hotfix is not yet available through Microsoft Update or WSUS, I think it is still important for people to understand the bug and its possible consequences, in order to identify it on unpatched systems:

veroblog.wordpress.com/.../excel-2007-bug-shows-wrong-answers-to-simple-multiplications

David Overton wrote re: Fix for Excel 2007 and Excel Services calculation errors
on Wed, Oct 10 2007 11:35 PM

Adam,

Its interesting.. if your car blows up while you are driving, there is quite a good chance you might die.  If Excel displays some numbers wrongly, but behind the scenes still thinks of them as the right numbers, then I would like to think no-one will die.  I think the bug was well and truly documented by others and MS did a lot of press interviews.  We now have a number of hotfixes out, so hopefully things will smooth out.  

I appreciate you concerns about more publicity, but I'm not sure who / how you hoped to impact by the information.  I'm not aware of many software houses or partners who declare all their problems and mistakes and shout from all the hills until they understand it and have a fix.  What use would it be to say "there is a problem", but not be able to say exactly what it is, why it occurs and how to mitigate it?

thanks

David

AdamV wrote re: Fix for Excel 2007 and Excel Services calculation errors
on Thu, Oct 11 2007 8:57 AM

I received a letter from Vauxhall when my Vectra had a problem which meant that the handbrake cable _might_ over time stretch slightly and mean that I had to pull harder on the handle to make it effective. I never found this problem personally, but I took it along anyway (a bit like accepting MS updates without having suffered from a problem, "just in case"). I was provided with a courtesy car while mine was in being fixed, at no cost to me, and for a problem I would almost certainly have never noticed. I am pretty sure my life was not in danger, this was a matter of QA.

To a brokerage firm (for example), miscalculating a transaction would not only cause loss of faith by their clients, but possibly substantial fines and sanctions by their regulatory bodies. To whom would they turn for compensation when they find that they had not been pro-actively informed of a known bug?

I would have thought that somewhere on the main Microsoft.com website in the Office section or the Excel pages here:

office.microsoft.com/.../FX100487621033.aspx

might have been a small link at least. "There may be a bug with some very specific calculations - see <link> for more details".

I'm not suggesting it should have been on the homepage, or plastered across the front of the campus buildings, nor even a mailshot to all users of the software(if you could even identify those). Just a bit more presence so that someone with sufficient interest in upgrading to have been reading the relevant sections of the website could make an informed decision.

The usefulness of this would be twofold - people could make better choices (including the choice that this was not a big issue for them if they felt that way), and people would see that MS is commited not just to more secure software but to better quality all round. "Trustworthy" (as in T. Computing Initiative) has many subtle shades of meaning, "having integrity" being one of them. Indeed, to quote from the TWC homepage:

"Reliability - We [MSFT] are committed to improving the quality of the technologies, products, and processes that customers need for systems that are reliable and that perform as intended and expected.

Business Practices - We strive to maintain the highest standards in our business conduct, to ensure integrity and transparency in all of our business practices, and to address society's ethical, legal, and commercial expectations."

I think it was clear enough what the problem was, and several people documented examples of when it occured, and which functions were most adversely affected by it. As to why it occurred - users should not need to care (and even now the fix is out, we are not being told why it happened, so this is rather a moot point). While it would be nice to think such a bug would not get through QA I can accept that such things can happen with a codebase this large and so many radical changes to core components (such as for multi-core calculations). The fact that the bug exists is not the issue here, that's just something to be dealt with as with any system error.

The best mitigation for anyone who did not already have 2007 would have been to continue with their current version until a fix was announced. This might have delayed some rollout plans by a couple of weeks at most, but at least anyone rolling out would have had the choice, and could plan for incorporating the hotfix as well (via WSUS or slipstreamed into a build, or using SCCM etc).

I happen to subscribe to the RSS feed for the Excel team blog on MSDN. I suspect 99% of Excel users do not.

Other IT support workers, power-users of Excel or MS partners selling or supporting MS Office might subscribe to blogs such as Darren's or the "Microsoft at Work" blog so they can hear about information which is useful in their discussions with users and clients. To find that this did not even merit a mention is frustrating.

I'm not angry, or upset, just mildly disappointed - the sort of feeling you would get when you find a broken vase hidden behind the sofa which your teenager did not come and tell you about but waited for you to discover. Telling you that they did own up by writing an entry in their diary is hardly going to make you feel that this was real "transparency", is it?

David Overton wrote re: Fix for Excel 2007 and Excel Services calculation errors
on Thu, Oct 11 2007 7:55 PM

Adam,

Did Vauxhall send you a letter before they had diagnosed the issue and had a plan on how to fix it, or did they wait a little while and understand the problem, then write about it to their dealers, put a fix plan in place etc?

thanks

David

David Overton wrote re: Fix for Excel 2007 and Excel Services calculation errors
on Thu, Oct 11 2007 9:17 PM

Just to add to this, what would you expect to be said before it had been diagnosed and fixed ... "There is a problem with Excel, sometimes it gives a wrong number... not 100% sure of all the cases when it does and we have no fix, but be careful and check all numbers with your calculator just to be sure?"

Of course, once we had diagnosed the issue fully it could have been "Excel goes wrong in these circumstances, if you have a calculation that might display a number like this using multiplication then please double check with a calculator as we have no fix at this time" and then shortly after "download this fix to stop some numbers being wrongly displayed".

I can see why you might have want the 2nd statement, but really, #3 is the only viable one in my opinion.  If you remember the Intel number processing bug, option #3 was the only one that was appreciated at the time, even though people spent a lot of time on working out what the bug was and writing solutions to not use the FPU, to most, until a fix was available, the information was not usable.

just my opinion.

Chaitanya Sagar, Excel Expert wrote re: Fix for Excel 2007 and Excel Services calculation errors
on Tue, Apr 7 2009 1:02 PM

Is this error only specific to MS office 2007.I first of all I don't understand why didn't Microsoft check this error before. Someone must have complained them about this error, at least they put up some update of the file which corrects the error. may i don't understand the complexities involved in it but i feel its the responsibility of the company set things right.

Chaitanya Sagar, Excel Expert wrote re: Fix for Excel 2007 and Excel Services calculation errors
on Thu, Apr 9 2009 6:55 AM

I think your explanation of the source of the problem is not clear. You are emphasizing that Excel is performing the calculation properly, but not explaining what component of Excel that is causing the results to be displayed incorrectly.

David Overton wrote re: Fix for Excel 2007 and Excel Services calculation errors
on Thu, Apr 9 2009 7:36 AM

Chaitanya,

The problem was a display only error, so it was the "display the number in the cell" part of Excel that was failing.  The service packs since this was known about include this fix, as well as the hotfix.

Add a Comment

(optional)  
(optional)
(required)  
Remember Me?

(c)David Overton 2006-20