Wednesday, January 9, 2008

Zen and the Art of MS Office Problem Determination

Is Software Problem Determination a Logical or Scientific process based on cause analysis and repeatable observation or is it a Zen style art, which has nothing to do with Western culture logic? 
I asked myself this question few days ago while trying to find why Microsoft's Office 2003 Word and Outlook stopped functioning. The answer is that maybe it is not Zen art training in which the user should experience unification of himself and the computer like the unification of the archer and the arch as a mean for hiting the target with an arrow in Zen Archery mastering. Surely, it has nothing to do with logical Problem-Determination. 
My background is different from the background of a typical home Office user. In the 1970s and 1980s I was an MVS Systetms Programmer responsible for the operating system in a site with few thousands users. My responsibilities include maintaining changes to operating system components in the Object code format and source code format (The programming language used was Asssembler). 
After you read the details rekating to my attempts to solve the problem, try to imagine how a typical user, with no IT background would address the same problem. 

The Problem 

While editing a Microsoft Word document, I received the following message: "Microsoft Visual C++ runtime library, A buffer overrun detected file microsoft office/office11 winword.exe A buffer overrun has been detected wich has corrupted the programs internal state. The program cannot safely continue execution and must now be terminated. Word was shutdown". 
Trying to open it again for the same file or most of the other files resulted in a menu of the Automatic Restore feature asking for deciding which file to save (In most of the cases the file has nothing to do with the optional Automatic Restore file names and content). Unlike many other problem instances, no probelm reporting to Microsoft dialog was presented. It should be noted, that this is the first time I am experiencing that phenomenon. 
An identical error message appears when I tried to send new Outlook messages and it was terminated same as Word (After the problem was partially solved I thought that it is due to the use of Word as the default editor by Outlook. 
During Problem Analysis I thought that a the error is within a component used by Word and by Outlook). According to Murphy's Law it occurs when I had to complete a lot of urgent assignmnets requiring Word documents and Outlook messages.

Trying to solve the problem


Step 1: trying to analyze the problem  

What caused the problem? The error message does not include possible error causes. Analyzing the problem cause is a tough mission for typical user, who does not know what buffer overrun is. I thought of the following possibilities: 
Possible Cause 1: Temporary bug in Office due to low probability state 
1. Open again the same document
2. Open another Word document
3. Reboot

Result: None of the actions solved the problem.


Possible Cause 2: Error within the document due to hahrdware or software problem

Action: Open another Word document

Result: The problem was repeated as before.


Possible Cause 3: Known bug

Action: Windows Update

Result: only two non relevant updtaes were found and installed. Nothing was changed as fas as the problem was concerned.


Possible Cause 4: A Trojan Horse or a Worm or a Virus

Action: Scan the disk for viruses

Result: No virus was found after scaning the whole disk.


Possible Cause 5: A Macro error

Most of the documents did not use macros.


Possible Cause 6: A change to the system caused the bug

Action: System Restore in order to take out the change

Result: restoring to two different restore points without any affect on Word's behaviour.


Step 2: Wondering what happened  

Probably, I was facing more sever problem: an unknown bug with no fix, an unknown Malware (Trojan Horse, Worm, or Virus). Knowing that most Anti-Virus solutions address new attacks only after they occur, I thought of the possibility that I am the unlucky vicitm this time: My anti-virus supplier will solve the problem after all my data will be destroyed. 


Step 3: A non-analytic Informatio Gathering  

Trying to find something by Google Search of the following : Word2003 'buffer overrun" Vista. The results revealed that it is quiet frequent phenemenon. I also found many occurences and many solutions in a forum discussion. Let us look at some examples (part of the web pages quoted here, for complete discussion please refer to the hyperlinked URLs).

First entry example

"This is a new laptop running Windows Vista with which I had no problems before installing Updates. However, I did not have SP3 on the previous laptop. All of this is in Spanish (Spain) versions but the message is in English. I tried turning off spell check but it has no impact. Spell check is essential for me anyway as I must write in 5 different languages." "I get the same error on mixed language files im my case (Hebrew and English) I have not found any com add ins. But by applying the remove spell check work around was able to load the mixed language file. This is a new phenomenon after installing sp3"

Second Entry Example

"I found another thread on a Hebrew forum where the problem appears to much more prevalent. One of the recommendations (other than uninstalling sp3) was to remove the custom dictionary. That worked and allowed spell check to run as well."

Third entry example

"> Lately I experience Buffer Overruns with Word 2003 documents with Office > 2003 NL Service Pack 2. This error never occurred, before the Service Pack > 2 > was installed. Word is in Dutch interface with Dutch Spelling and Grammar. > > Exact message: > < > Microsoft Visual C++ Runtime Library > > Buffer overrun detected! > > Program: Files\Microsoft Office\OFFICE11\WINWORD.EXE > > A buffer overrun has been detected which has corrupted the program's > internal state. The program cannot safely continue execution and must > now be terminated. >> > > This is a very annoying error because Word 2003 always terminates and > there's no possible way to edit the documents. > So I edited some of the documents with Office XP UK on an older computer > with help from a friend. After some trial and error, I found out that when > you have > Spelling and Grammar disabled in a TextFrame with some lines of text, this > error doesn't occur anymore. > Under menu: Tools -> Language -> Set Language -> Do not check spelling or > grammar. > > > I have several questions: > Is this already known as a bug? > Where can I find a fix for it, because I don't want to edit all my > documents > in Word XP, just to be able to open them in Word 2003?" "TF" wrote: > I've not seen or heard of this one. It may, of course, be a Dutch specific > bug. However, have you tried running a Detect and Repair operation (under > the Help Menu in any of the Office apps)? > > -- > Terry Farrell - Word MVP > "


Step 4: Which Solution would you choose?  

It is clear that it happens when using mixed languages text (English and any other language such as Dutch , Spanish and Hebrew) Would you reinstall office? Would you ask for a friend's help to copy the file to his PC including another Windows Operating System version? Would you disable a dictionary or a speller? I thought that trusting Microsoft's MVP is the best and easiest (see the red highlighted part in the third entry example): Detect and Repair in the Help menu. However, without any previous warning this method scratched all my Outlook messages and contacts.

Step 5: What is Detect and Repair? 

I thought Detect and Repair tool should locate and correct the problem in the C++ library d. However, Detect and Repair, detects nothing and repairs nothing, it reinstalled the Office sub product again (No warning message was send prior to this action execution). As far as Word is concerned this unexpected behaviour cause no damage, but as a result of tagging Outlook I lost all my messages and my Contact list.  

Imaginative scenario: Average PC User  

Average non-IT user would try in vain to delete the document or reboot his computer. His next step would probably be taking his computer to a store or a laboratory and asking an expert to help him (or call an expert for home visit). Some of these experts would reinstall Office. Others may search in the Web and find temporary circumvention. Reinstalling Office may result in identical call for expert help few days after (or may be not, if the new installation will reset a field containing a wrong value). Unhappy End and unanswered questions After being led down by the MVP's advice I tried the non-expert advice and turned off spell check which solves the problem in 90% of the cases.

Some of the mystical unanswered questions are:

  • Why that problem does not occur to me previously? For few months I am writing and editing successfully a large number of Dual Language Word documents using the same system. 
  • Why no Microsoft update is available for solving the problem? Some of the posts are from November 2007. 
  • Why no dialog-box for sending problems details was available? Even if the details will not be too useful for Problem Determination, at least the frequency will be known.
  • If a warning dialog box appears so often in Windows Vista as a Security measure and before other operations such as deleting a file, is it reasonable that before deleting all Outlook Messages and Contacts no such message appears in "Detect and Repair" operation? 
  • Does expertise matters?


Anonymous said...

Thank you for your blog entry, which helped clarify the problem for me. I resolved it by resetting the language defaults in Word. If anyone wants to do this, go to Tools, then Language, then Set Language and reset the default in each document to English. Now the documents don't crash at all. I still don't understand why that fixed the problem, but it did.

Anonymous said...

Thank you for the blog
I am stuck with the same problem.
Intalled XP SP3 and Office 2003 XP3 did not solve the problem.

I will try spybot to see if any worms are here. symantec AV did not find any virus

Dov Grobgeld said...

My wife encountered the same problem, and thanks to your blog, we were able to workaround it. Thanks!

This is an example of the danger of closed source. Except for staring at the meaningless error message and trying lots of black box scenarios, my hands are tied behind my back. Had I had the source code (and preferably running under Linux), I would already launched my debugger.

Alexis said...

Some days my friend called me up and told me about the interesting tools. But one of its liked me and I used for recovering my emails - outlook database repair. It surprised me,because of restored my corrupted emails for seconds and without payment as I bore in mind. I recommended it my friends and they thanked me a lot.

Public Cloud Core Banking: Hype or Reality? - Revisited

  More than 4 years ago I was asked if Public Cloud Core Banking is a Hype or a Short Term Reality? If you had read the post, you would prob...