Re: Strange behavior of TBranch::Fill

Dear Philippe,

   thanks for your reply.

The problem is that in my case I really don't need to have all branches filled with the same number of entries. The example I sent around was oversimplified: in my real life case I have several branches, none of them is empty of course, but they for sure have different number of entries, at least this is what I would like to have!

I will try to find another solution....

thanks again!

Cheers,
Antonio

-- 
Antonio Bulgheroni, PhD

Computers have a lot in common with air conditioners:
Once you open WINDOWS, they stop working properly.

On 7/19/07, Philippe Canal <pcanal_at_fnal.gov> wrote:


>

> Hi Antonio,

>

> This is the expected behavior of your code.  In you branch->Fill case, the

> tree is never informed that it has any entries ... and hence the tools

> (TBrowser, etc..) thinks the tree is empty.  For most of the tools, it is

> important that each branch to be filled with the

> same amount of entries (because they match entries by their sequential

> order

> in the branch).

>

> You can 'fix' a TTree after being filled by your branch->Fill method by

> calling 'tree->SetEntries()' but it will complain if one of the branch has

> less entries than the other.

>

> You can also use SetBranchStatus(brname,0) and tree->Fill to prevent the

> filling of individual branches.

>

> > I don't need (and don't want) to fill all branches,but I would like to

> indivually fill only branches I'm interested in.

>

> Why?  What do you want to represent with the 'empty' branch?  Wouldn't it

> be

> better to __not__ add the branch to the TTree.

>

> You may want to consider replacing the

>       TBranch * branch2 = tree->Branch("calib2.", &calib2, 8000, 1);

> by

>     vector<Calib> *vec;

>       TBranch * branch2 = tree->Branch("calib2.", &vec, 8000, 1);

> where the vector would be filled or not.

>

> You may also want to replace both branch with a single vector with one of

> more object in it,

> depending on the semantic.

>

> Cheers,

> Philippe.

>

> ________________________________

>

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

> On Behalf Of Antonio Bulgheroni

> Sent: Thursday, July 19, 2007 1:55 AM

> To: ROOT Talk

> Subject: [ROOT] Strange behavior of TBranch::Fill

>

>

> Dear ROOTers,

>

> I would like to report you the strange behavior of the TBranch::Fill

> compared to the TTree::Fill. In my real life case I have a TTree with

> several branches; I don't need (and don't want) to fill all branches,but I

> would like to indivually fill only branches I'm interested in. For this

> reason, I'm getting from the TTree the branch I need, I'm setting the

> address of the TBranch to a local variable and finally I call

> TBranch::Fill(). Everything is working fine but the tree browser, indeed

> if

> I browse the TTree and double click on a leaf instead of drawing the leaf

> content on a canvas, nothing is happening.

>

> I discovered that if I replace the TBranch::Fill call with a TTree::Fill,

> the tree broswer is working again, but every time I have

> all branches filled. Here attached a simplified example, calib.h and

> calib.cxx contain the class defining the branch content, and test.C a

> simple

> macro to fill in a tree using in one case TBranch::Fill and in the other

> TTree::Fill.

>

> Is this the correct behavior? If so, is there any workaround to have

> branches browsable and individually fillable? I played a bit with the

> branch status, but unsuccessfully.

>

> Thanks in advance.

>

> Cheers,

> Antonio

>

> Tested on the following configurations:

>

> ROOT 5.15/08 + Linux (Fedora 7) +  g++  4.1.2

> ROOT 5.13/04 + SLC 4 + g++  3.4.6

> ROOT 5.10/00e + SLC 3

> --

> Antonio Bulgheroni, PhD

>

> Computers have a lot in common with air conditioners:

> Once you open WINDOWS, they stop working properly.

>

>