Werner Exinger wrote:
StuartBruff schrieb:
Did you play with the exclusion range as the instructions at the end of the recovery worksheet? I've just had a quick go and found that excluding region number 37 (region-id 328) restored the rest of the worksheet (that is, I set k in the purple region to 37..37, which then cut that region out of the worksheet). As I said, it involves rather too much user interaction to appeal to my tastes. I was hoping to find a way to automate the region exclusion process but didn't manage to do so.
(Having looked at the offending region, I opened up the original worksheet in Notepad++ and manually deleted the section between lines 805 (result) and 842 (/result); that restored the worksheet as well)
Stuart
No, I didn't play with the region exclusion, as I had no idea where to start. How did you found out (quickly) that region 37 was the faulty one?
Manually deleting that whole result-section was exactly what I did in first place.
First thing, which I gather you did, is to change the variable FileName in the magenta-highlighted region to your desired filename, excluding the ".xmcd" extension.
The worksheet will (should!) then read in the xmcd file and copy all of the regions into the vector regns.
The next step is to look at the number of regions converted (rows(regns)=), which gives you an idea of how many regions you've got to analyze. Having ascertained this, you set the vector exclusions to list the regions you think may be bad or that bracket the bad region. The simplest way is to assume that any bad regions lie in a contiguous block (IME, this is usually the case), then set the range variable to k to cover the last half of the regions (eg, if there are 80 regions, set k:= 40..79); this will then reconstruct the original xmcd file using only the first half of the regions. Open the resulting worksheet and see if it runs OK.
If it doesn't, drop k to, say, 20 and try again. Keep reducing k until the worksheet runs ... if it still doesn't, then the problem doesn't lie in the regions and the recovery worksheet won't help you.
If it does work, then either increase the lower bound of k (eg, 60..79) or decrease it's upper bound (eg, 40..59 or 20..40) and try again. Repeat until you've found the offending block of regions. By using a binary search pattern, the problem can usually be found fairly quickly.
Occasionally, you may be unlucky and have two or more widely separated problem regions, in which case iteratively hand modify the exclusion vector (using, say, stack) to identify the problem regions. I've only had to do this a couple of times.
As you can hopefully see in the image below, I'd identified the 37th region (region-id 328) as the problem and after only about 5 iterations. I then had a look at the region, couldn't see anything too obvious but just had a play starting with the 'result' section as that contained an error description. Et, Voilà!
Stuart
