Re: [pdf-devel] LZW filter implementation

[jemarch]

Hi JP.

   I have finished the LZW implementation. 

Thanks.

   Two notes:

     - With cache < 5 we cannot assure that some characters are missed
       in a compression-decompression cycle (footer sequence sequence
       takes several chars).

       [...]

       I would have fixed this but with the previous issue (which is an
       architecture problem, not implementation).


I dont understand the reason for this limitation. It is related to a
problem onn the streams architecture or with a problem on the filter
internal architecture.

A minimum size of 5 for the stream cache is reasonable, but if it is a
problem in the streams implementation more problems may arise in the
future.

   BTW, I am very busy for the rest of October. Is it too much to wait for
   the predictor's reimplementation till the first week of November?

It is ok, many thanks.