Re: [Bug #2955] .idata file becomes zero size if disk becomes full

Jason Rennie <[email protected]>:

> [email protected] said:
>> Hi.  Here's a request to reactivate this bug and test for a race
>> condition, my .idata got nuked by ifile processing recently.
>> (similarly to the mailing list post, my user directory partition was
>> temporarily overflowing and ifile wrote nothing back to .idata.) 
>
> Are you using version 1.3.1? ("ifile --version" will tell you)

i don't know what ifile does in situations like this, but the
"standard" way of doing this would be to write a temporary database
onto the same filesystem and only rename(2) it into its final state
after the update has been successful.  if not, it might return
"temprary error".

  clemens