When using zlib (DEFLATE), do you know cases where using full flushes
perform better than partial flushes?
I ask because RFC 4978 (IMAP COMPRESS Extension)
A server can improve downstream compression if it hints to the
compressor that the data type is about to change strongly, e.g., by
sending a Z_FULL_FLUSH at the start and end of large non-text
I wonder whether it is really the case.
When using Z_FULL_FLUSH, the sliding dictionary of the previous 32K
bytes of uncompressed data is wiped clean. So I am under the impression
that this will always result in compression that is less effective than
or equal to doing a Z_PARTIAL_FLUSH. Having the 32K dictionary from
previous data, no matter how unrelated, can only help; and if there are
no matching strings in there, then it is simply not referenced.
Could it be that when using Z_FULL_FLUSH, zlib would start again from
scratch and therefore use <15 bit offsets for the backquotes, and use
fewer output bits to adapt to how base64 looks from scratch than with
10K or more of ASCII text experience?
Have you ever seen that? (a full flush improving compression)
Thanks beforehand for your comments on that subject.
« C'est une liqueur de grain distillée en Calédonie. » (Astérix)