Consider declaring dynamic data methods unconditionally about cld2

[GoogleCodeExporter]

[email protected] who raised this issue in Chromium code review 333603002:

https://codereview.chromium.org/333603002

Original comment by [email protected] on 23 Jun 2014 at 9:58

[GoogleCodeExporter]

This is looking more important now that Chromium is running into issues that 
exacerbate the problem:
https://code.google.com/p/chromium/issues/detail?id=367239

In a complex project that uses CLD2, having to fork the build logic on 
CLD2_DYNAMIC_MODE is sub-optimal.

I suggest we go ahead with the fix proposed above, with one caveat: While we're 
here, we add a new method:

  bool isDataDynamic()

This method can be declared in the .h and implemented with a simple 
preprocessor check in the c++ code to return either true (if CLD2_DYNAMIC_MODE) 
or false (otherwise). This will still allow dependent projects to be smart. The 
cost is that we will need to bring in a total of 5 function signatures into the 
header (the 4 that already exist, plus the new one); this shouldn't be a big 
deal, though, since there's no extra includes and build optimizations should be 
able to simply discard the unused symbols in most projects. Very low cost, and 
makes life easier. I'm going to go ahead with this.

Original comment by [email protected] on 25 Jul 2014 at 8:07

[GoogleCodeExporter]

Committed in r162.

Original comment by [email protected] on 25 Jul 2014 at 9:56

• Changed state: Fixed