Re: Re: untar with plzip faster than with lzip for no apparent reason

[wrotycz]

In a lzip decompressor, dictionary and output buffer are the same thing.

Makes sense, but still, you should be able to output it more often. Plzip can.

Also when the dictionary is full but data block is no reset, still dependant on previous data (here: if you pass 32 MB decompressed, as if input was 100MB and single member) you have to maintain 32 MB dictionary and get rid off constantly 'obsolete' data, beyond 32 MB range.

If you find the lzip source a bit complicated, you can check it in the lzip specification[1].

How many times do I say I read it?

The dictionary buffer is allocated in the constructor of 'class LZ_decoder' like this 

  buffer( new uint8_t[dictionary_size] ),

and is the only buffer used by the decompressor.

Will check.