Re: [Lzip-bug] LZ_decompress_finished before LZ_decompress_read

[Antonio Diaz Diaz]

Hello Jacob,

Jacob Rief wrote:
> currently I am adding LZMA support to an existing streaming library
> using your lzlib implementation.

Great!


> this is OK, but LZ_decompress_finished only returns 1 after
> LZ_decompress_read has returned with -1, setting LZ_decompress_errno
> to LZ_header_error.

Are you feeding any trailing garbage to LZ_decompress_write? Lzlib can 
decompress concatenated lzip members (compressed data sets). So, if it 
finds more data after the end of a member, it will try to decode a new 
header.


> This implementation works fine, but from the point of view of an API
> client, I would expect for cleanness and symmetry reasons that
> LZ_decompress_finished returns 1 BEFORE LZ_decompress_read
> returns an error.

This is the intended behaviour of lzlib in absence of trailing garbage.


Best regards,
Antonio.