[moneydance] data file corruption

[Brett Russ]

On 4/25/07, Gordon Alley <gordon at gordonalley.com> wrote:
> I just wanted to relate this incident to the list in case it's not just me.
<snip>
> When I tried to open my MD file, MD asked me if I'd like to open the
> more-recent backup file, and I told it to do so. But, to my chagrin,
> after a short delay I was presented with an alert that said "Error:
> Cannot read file: java.io.EOFException". Also, when I look at the
> files, the backup is about half the size of yesterday's file. Doesn't
> look good!

I had pretty much the exact same thing happen yesterday.  I am not
sure what 2007 build I was running, but it was fairly recent.  My
Linux machine froze and I needed to hard power cycle it (!!) to get it
back.  When I launched MD it prompted me to a) upgrade to 566; and b)
use the autosave file.  I said yes to both.  The autosave file gave
the same exception you noted and had 0 size.  The main file was dated
2007-04-21, or 3 days old.  Therefore I lost 3 days of manually
entered transactions which I swore I had manually saved at least once
a day.

No problem I thought, I have MD set to do daily backups.  I checked my
backup directory and saw the most recent backup was also 3 days old!
And there were only 5 backups in that directory despite me specifying
to keep unlimited backup files.

So...couple problems here:
1) autosave file empty or at least corrupted
2) daily backups don't happen (this is big)
3) daily backups are pruned when they shouldn't be

Can someone on the development team describe at a high level how the
autosave and backup code works?  How could backups have malfunctioned
for 3 days?  How could autosave write garbage?  Could there be more
checking of sane autosave and backup files in the code?

Thanks,
BR