Dictionary entry vs. compiling & loading class

ROOT
#1

Hi,

I have a class (MyClass.cpp) that I want to load and use in a macro (my_macro.C). What is the difference between (a) compiling & loading the class to memory and (b) making a dictionary entry?

i.e.

(a)
gROOT->ProcessLine(".L MyClass.cpp+");
gROOT->ProcessLine(".x my_macro.C+");

vs.

(b)
including
#pragma link C++ class MyClass+;

in the macro and then simply running
gROOT->ProcessLine(".x my_macro.C+");

Edit: Let me make myself a bit more clear. In (a), I get a MyClass.so file; in (b) I don’t. In (b) the execution of my_macro starts much faster than in (a). I was wondering if that means that the code in (b) is not optimized(?) and/or if it runs slower than in (a).

Thanks.

–Christos

#2

Hi,

There are 2 main differences.

In your case a), it exposes to CINT all the symbols declared in MyClass.cpp and MyClass.h, my_macro.C (and my_macro.h).

In b), it exposes to CINT ‘only’ the class MyClass, and the symbols in my_macro.C (and my_macro.h).

Also in a) you generate 2 dictionaries and compile/link 2 separate shared library.
In b) you generate only 1 dictionary and compile/link onlu 1 shared library.

So in conclusion b) is faster but possibly do not expose as much to CINT.

Cheers,
Philippe.

#3

Hi Philippe,

So, the startup time is smaller in (b). What about the execution time? Are (a) and (b) equivalent?

–Christos

#4

The execution time should be the same.

Cheers,
Philippe.