Re: Changes to allow compiling with -Werror using gcc 4.6.2 on i386. (issue 5489092)

[md5i . mail]

http://codereview.appspot.com/5489092/diff/1/lily/beaming-pattern.cc
File lily/beaming-pattern.cc (right):

http://codereview.appspot.com/5489092/diff/1/lily/beaming-pattern.cc#newcode187
lily/beaming-pattern.cc:187: I64 count = 1;  //default -- 1 base moments
in a beam
On 2011/12/20 01:54:29, Graham Percival wrote:
> huh, I've never encountered I64 before... why does this particular int
need to
> be explicitly 64 bits?  If we really can't fit that data into 32 bits,
it would
> be nice to add a comment explaining why.
The variable count has to be I64, because it may be set to the result of
test_count, which is I64.  test_count is I64 because tuplet_count is
I64, and tuplet_count is I64 because factor.num() is I64.  In each case,
to use a smaller width could, in theory as far as the compiler is
concerned, truncate a value.  In practice, this almost certainly never
happens.  The alternative is to explicitly cast I64 values to int.  That
may be a better approach, but I prefer to take a conservative approach,
and avoid casts whenever possible.

http://codereview.appspot.com/5489092/