Re: [Lzip-bug] Making lzip and plzip interchangeably usable

[Antonio Diaz Diaz]

Hello Daniel.

Daniel Baumann wrote:
> it would be nice if lzip and plzip would be 'drop in replacements for 
> each other' so that i could install plzip and have lzip symlinked to 
> plzip, that way all my stuff on the system would use plzip without 
> changing anything.

They are already "drop in replacements for each other"... as long as one 
sticks to common options. :-)


> for this to work, the lzip and plzip would need to share the same 
> commandline parameters. fortunately, they almost do.. the only 
> differences wrt/ current releases are:

Lets see how those can be harmonized.


> lzip:  -b, --member-size=<bytes>      set member size limit in bytes
> plzip: -B, --data-size=<bytes>        set input data block size in bytes
>
> i guess this is supposed to control the same thing, no?

Not at all. They are in some sense the opposite of each other.

Plzip splits the input file into chunks and then compresses each chunk 
independently, producing members of variable size depending on the 
compressibility of input data.

Lzip compresses all the input file as a single stream but is able to 
produce multi-member output with members of (about) the same size. The 
amount of uncompressed data in each member is variable, depending on the 
compressibility of input data.

It is not possible for plzip to implement -b (plzip currently ignores it).

It would slow down lzip to implement -B, and I am not sure if it is a 
good idea to implement -B in lzip as a noop, given that the default 
value is usually good enough for plzip.


> plzip: -n, --threads=<n>       set the number of (de)compression threads
>
> lzip doesn't have a -n option, it would be good if it had one and would 
> just ignore it (noop).

Yes. This is the easiest to "fix".


> lzip:  -S, --volume-size=<bytes>      set volume size limit in bytes
>
> i guess the right thing would be to add support for volume size limits 
> in plzip.

It could be "the right thing" if it were possible to do it. (plzip 
currently ignores it).


> lzip:   0 .. -9                       set compression level [default 6]
> plzip: -1 .. -9                       set compression level [default 6]
>
> plzip should start at '0' instead of '1' for consistency anyway.

Level 0 is special in lzip. It uses a fast encoder not present in plzip. 
Plzip currently accepts -0 as a synonym of -1.


> all but the --volume-size are trivial things, would you accept patches 
> for that?

The only easy one that is not already done is -n. I'll make lzip 1.15 to 
ignore it. (it is late for lzip 1.14).


Regards,
Antonio.