Re: rootcint generates uncompilable code if no LinkDef.h "+"

  Hi Robert,

 > Is there some reference for when _not_ to add the "+"?  I'd like to
 > sweep through the two projects and add "+" (almost) everywhere but
 > I'd also like to avoid breaking anything that might arise from being
 > overzealous.

See the user's Guide on some issue when moving from 'no plus' to '+' for persistent classes (See the User's Guide's chaper on I/O for details) for backward compatibility.

Also note that for classes __without__ a ClassDef, changing from 'no plus' to '+' changes from no-i/o to with i/o.

Cheers,
Philippe.

On 9/7/10 3:19 PM, Robert W. Hatcher wrote:
> Recently with the SVN HEAD of ROOT (on Scientific Linux 4.7 w/ gcc 3.4.3)
> rootcint has been generating code that won't compile for classes
> that have a STL map where the primary key value is an enum.
>
> Attached is a tar.gz file w/ an example drawn from one of two separate
> projects that has gotten bitten by this. Attempting to compile the
> dictionary source gives:
>
> _ROOT_DICT_pkg2.cc: In member function `virtual void myspace::MyClass::Streamer(TBuffer&)':
> _ROOT_DICT_pkg2.cc:294: error: cannot convert `const myspace::MyEnum_t' to `void*' in initialization
> _ROOT_DICT_pkg2.cc:294: error: expected `,' or `;' before '=' token
> _ROOT_DICT_pkg2.cc:295: error: `ptr_' cannot be used as a function
> Unwind the tar.gz file, cd root_rcintmap, and ./makescript.sh to
> demonstrate this problem.
>
> Adding a "+" to the LinkDef.h line for the class with the map<enum,x>
> (i.e. MyClass) resolves the problem. I see that some of the
> documentation on LinkDef.h says:
>
> / Note the "+" after the class name:/ this enables an essential feature
> for rootcint. We would love to make it the default but in some cases
> we will break code - so we have to ask you to add that "+" at the end!
>
> Is there some reference for when _not_ to add the "+"? I'd like to
> sweep through the two projects and add "+" (almost) everywhere but
> I'd also like to avoid breaking anything that might arise from being
> overzealous.
>
> -robert
>
> Robert W. Hatcher | rhatcher_at_fnal.gov <mailto:rhatcher_at_fnal.gov> 630-840-3102
> FNAL CD/REX (MINOS) | MS 220, PO Box 500, Batavia IL 60510
>
>
>
Received on Tue Sep 07 2010 - 22:36:49 CEST