[Carpet] [Carpet-darcs] CarpetIOASCII: added utility script to merge CarpetIOASCII output written before and after recovery

[Thomas Radke]

Erik Schnetter wrote:
> On Aug 15, 2006, at 10:10:15, Thomas Radke wrote:
> 
>> Erik Schnetter wrote:
>>
>>> The largest problems are datasets which are only partially written,
>>> because the process was interrupted (or crashed because it ran out of
>>> disk space).  That can lead to only parts of lines being present.  I
>>> would be nice if this could be handled as well.
>>
>>
>> I can try. Do you have any such output file(s) ?
> 
> 
> Not at the moment.  One way of generating such a file is taking an  
> existing file and truncating it after e.g. 10240 bytes or so.

So you want to keep the newest dataset and discard older ones ? This 
isn't possible when reading from a stream - the script would then need 
to read from a (set of) physical file(s) in two stages: (1) to analyse 
all the datasets and sort them, and (2) to actually output them again, 
omitting duplicates.

Anyone has a better idea ?

-- 
Cheers, Thomas.