One of the main limitations of the LZ77 algorithm is that it uses
only a small window into previously seen text, which means it continuously throws away
valuable dictionary entries because they slide out of the dictionary. The longest match
possible is roughly the size of the look-ahead buffer; many of the matches actually
existing in the file may actually be much longer.
The sliding window makes the algorithm biased toward exploiting recency in the text,
and this is not necessarily a loss. This can be a significant advantage, e.g. in a
phonebook streams of data tend "look" more like what has been seen recently that
what was seen long ago." (Nelson, 277) However, the street address would only show
compression faintly, when two listings with the same last name lived at the same address.
One obvious way to get around this problem is to increase the size of the search and
look-ahead buffers. However, changing these parameters will drastically increase the CPU
time needed for compression. Since string comparisons between the search buffer phrases
and the look-ahead buffer proceed sequentially, the runtime here will increase in direct
proportion to the length of the look-ahead buffer. Also, changing the size of the search
buffer will only marginally affect the efficiency because we are still using an algorithm
that relies on recency to perform compression.
