Re: [Lzip-bug] lziprecover

[Antonio Diaz Diaz]

Hello Ole.

Ole Tange wrote:
> I was puzzled that nothing happened and 'lziprecover -v' was just
> sitting there. That is until I did 'ls'. Oh my freaking god! 58000
> files.

I recognize I didn't even think about so many members in a file.


> I even got: newton.log.lzip lziprecover: too many members in file
> (there seems to be a limit at 100000 parts - is there any reason for
> that?).

Yes. It is the same limit bzip2recover uses. :-)

As I see the current limit is not large enough, I'll implement an 
improvement I had already planned. I'll make lziprecover to use just the 
needed decimal positions for the number of members in the input file.

For example, a file with 6 members will use names like 'rec1<name>', 
while a file with 6 million members will use names like 'rec0000001<name>'.


> I had expected -v would be somewhat more verbose (e.g. telling me it
> was writing files).

Splitting files (with not so many members) is fast, so I didn't use -v 
on it. I'll make it more verbose now.

But another improvement I have planned is to repair (or merge) the files 
'in place', without having to split them before. Stay tuned. :-)


> And I would like an option to 'lzcat' the result instead of putting it
> into files.

I do not understand. Do you mean something like 'lziprecover -cd file.lz'?


Best regards,
Antonio.