Assume debug builds means it's a developer working on TigerVNC
rather than a user just compiling it. At that point we should not
accept any new warnings in the code.

Solve it by casting to the larger of the two, i.e. 64-bit.

Makes it more sane what type to use when interacting with this class.

Also avoids %lld which isn't supported on Windows.

The constants used here are unsigned so it makes more sense that
the parameters are as well.

Either by casting, or switching to a more appropriate type
for the variable.

This allows the compiler to analyse the format strings and
arguments for correctness.

The logic is a bit complex so we're going to have to silence the
warning with this explicit assignment.

Java viewer was incorrectly attempting to decode incoming clipboard updates to UTF-8 instead of Latin-1.

Updates to package builds

The system libpng is too old for fltk 1.3.3, resulting in undefined reference errors when trying to link vncviewer.

Commit f0fffd4d7f makes ldnow patch unnecessary.
Commit 0dc36a501e makes format-security patch unnecessary.

Static libraries are too hard to get linked correctly with cmake.

Don't violate RFB spec by sending UTF-8

Java viewer was violating the RFB specification by sending UTF-8 encoded clipboard text rather than ISO-8859-1.

This prevents unhandled exceptions if the file isn't found.

For 16-bit and 8-bit datasets, you can now run encperf with an argument of
-translate=0 to benchmark just the encoder without converting the datasets to
24-bit.  This allows the output of encperf to be directly compared with that
of the TurboVNC Benchmark Tools.