Re: TTree::Draw()/Scan() with deeply nested std::vectors

Ah, thanks for the info.

Are there any plans to lift this restriction? Is it extremely difficult to do so?

---
Stan Seibert


On Nov 15, 2005, at 10:02 AM, Philippe Canal wrote:



> Hi Stan,

>

> TTree::Draw can only only 2 levels of variable size arrays

> within an event (the equivalent of a total of 3 nested loops

> including the loop over the events).

>

> After your reach the 2nd variable size arrays within an

> event, the automatic unrolling of the container is

> diabled (because it would not be able to plot them anyway).

>

> The error message indicates that TTree::Draw has not

> automatically unrolled the 3rd vector and is looking for

> the next information (foo4) directly on the vector instead

> of its content.

>

> Cheers,

> Philippe.

>

> -----Original Message-----

> From: owner-roottalk_at_pcroot.cern.ch [mailto:owner- 

> roottalk_at_pcroot.cern.ch]

> On Behalf Of Stan Seibert

> Sent: Tuesday, November 15, 2005 9:26 AM

> To: roottalk_at_pcroot.cern.ch

> Subject: [ROOT] TTree::Draw()/Scan() with deeply nested std::vectors

>

> I recently stumbled on a problem using TTree::Draw()/TTree::Scan()

> with a TTree that contains a branch of objects with deeply nested

> vectors.

>

> For example, I have a branch in my TTree which contains C1 objects.

> Each C1 object contains a vector of C2 objects, and each C2 object

> contains a vector of C3 objects, and each C3 object contains a vector

> of C4 objects.  If I try to Draw() or Scan() a member variable in C1,

> C2, or C3, I get the expected behavior.  (That is, it iterates over

> all the nested lists down to the variable I have requested.)

>

> However, if I try to Draw() or Scan() a member of C4, I receive the

> following error:

>

> Error in <TTreeFormula::DefinedVariable>: foo4 is not a datamember of

> vector<C4>

>

> *ERROR 30 :

> Bad numerical expression : "c1.c2.c3.c4.foo4"

>

> This is curious, since foo4 is a member of C4, but clearly not a

> member of vector<C4>.  I can confirm this behavior on ROOT 4.04.02g

> and 5.06.00.

>

>

> I have attached a simplified script which reproduces this error.

> Since it uses vectors of user-defined objects, you have to run it

> with ACLiC:

>

> root nest.C+

>

> ---

> Stan Seibert

>

>

>

>