Write code update

Implemented better defaulting of zlib settings based on image properties.
Implemented pass-through of png_write_rows when the rows can be used directly (a
common case) optimizing the handling of previous-row buffering.

Removed the METHODICAL filter selection method and disabled the HEURISTIC one;
the first was ridiculously slow (though useful for experiments) the second
doesn't work.  Filter selection is temporarily disabled (it defaults to the
lowest numbered filter in the list; typically 'none').

New handling of compression settings (incomplete), new PNG compression level
(not yet visible in an API).

Back ported 'PNG_FAST_FILTERS' from 1.6 (in png.h).

There are minimal API changes beyond removal of the selection options.  Work is
still to be done to investigate a filter selection mechanism that is at least as
good as the previous one.

Signed-off-by: John Bowler <jbowler@acm.org>

scripts/pnglibconf.dfa

@@ -930,20 +930,36 @@ option CONVERT_tIME requires WRITE_ANCILLARY_CHUNKS
 #     png_set_filter interface allowing the application to select the filter
 #     used for each row.
 #
-#  SELECT_FILTER_HEURISTICALLY
-#     Enables code to cause libpng to choose a filter from a set passed to
-#     png_set_filter.  Without this code libpng just chooses the first filter in
-#     the list if multiple are given.
+#  SELECT_FILTER
+#     Enables code to select between multiple filters on write.  Without this
+#     the 'first' (lowest numbered) filter will be selected an this typically
+#     works out as PNG_FILTER_VALUE_NONE.
 #
-#  SELECT_FILTER_METHODICALLY
-#     Enables code to try all the filters in the list passed to png_set_filter
-#     and choose the one which results in the least number of compressed bytes
-#     added by the current row.
+#  COMPRESSION_BUFFER_MAX
+#     WARNING: take care if you set this.  This is the maximum amount of input
+#     data that the implementation of deflate can consume before it outputs a
+#     Huffman table for that data.  I.e. before it commits to an encoding of the
+#     data it has read.  This is used solely to implement a limit on the amount
+#     of image data buffering that occurs inside libpng before filter selection
+#     is done.  Normally the limit is never reached because of the next setting,
+#     but this is a compile time limit and it is intended to prevent a potential
+#     DNS service as a result of an application setting the libpng equivalent of
+#     volume level 11 (read the wikipedia article on "Up to eleven").
+#
+#     NOTE: the image of a black cat in a coal mine obviously requires this
+#     limit, but some more valid images can get very close; well over 8MByte.
+#
+#  COMPRESSION_BUFFER_LIMIT
+#     This is the (overrideable) default for the amount of memory libpng will
+#     buffer before selecting a filter for a row.  It is limited itself to
+#     COMPRESSION_BUFFER_MAX as values above that level make no change (see the
+#     previous paragraph.)
 #
 #  See png.h for more description of these options.
 option WRITE_FILTER requires WRITE
-option SELECT_FILTER_HEURISTICALLY requires WRITE_FILTER enables SET_OPTION
-option SELECT_FILTER_METHODICALLY requires WRITE_FILTER enables SET_OPTION
+option SELECT_FILTER requires WRITE_FILTER disabled
+setting COMPRESSION_BUFFER_MAX default 8453377
+setting COMPRESSION_BUFFER_LIMIT default 8453377
 
 # added at libpng-1.5.4