From serban@bnl.gov Thu Apr 29 16:24:27 2004 Date: Tue, 06 Apr 2004 15:48:43 -0400 From: Serban Protopopescu To: Yann Coadou Cc: Dugan ONeil , Gordon Watts , greenlee@fnal.gov, Serban PROTOPOPESCU , Yaroslav Kulik , deliot@hep.saclay.cea.fr Subject: Re: Trigger info and the data format working group [ The following text is in the "ISO-8859-1" character set. ] [ Your display is set for the "US-ASCII" character set. ] [ Some characters may be displayed incorrectly. ] Hi, I wouldn't say it has been decided by the group whether to go the tmb tree root way. What I am saying is that developing yet another format is not going to go anywhere. Perhaps I mispoke, I was under the impression that top root files are an ntuple format, not a tree with branches, TCloneArrays and the rest of the root tree paraphernalia. Serban Yann Coadou wrote: >Hi Serban > > > >>The way I see it there are basically two choices: >>1) you use classes and make your information available in object oriented >>form, in which case all we have right now is tmb root tree >>2) you fill a root tuple with the information in arrays. >> >> > >What Dugan meant here is that there are already many different ways of >producing different root trees. tmb_tree is one of them, you also have >the top_analyze version or the trigsimcert one, and I guess several >others. All those programs produce an object oriented root tree, but >with different structures/syntax. Since until now there was no official >format, we decided to create our own classes (based on what Josh Dyer >had done originally). Now it looks like the group that was appointed for >a common root format has as a charge not to find the better way, put to >expand on one that is already chosen, the tmb_tree format. Fine by me. >If this group had been formed 2 months ago, I would have put all my >effort directly in tmb_tree, but it was not the case then. Given the >time scale, we will finish trigsimcert, and I'll then be in touch with >Gordon to make tmb_tree really trigger useful. As far as I know there is >a lot of missing trigger stuff so far, and level 3 information is >definitely not user friendly. We can remedy this, it's just time >consuming and not my priority for the coming week. > >Cheers, >Yann > > > >>Dugan ONeil wrote: >> >> >> >>>Hi Gordon, >>>OK. We did look at the code in TMBTree before starting. The reason that >>>we didn't just add things to TMBTrees is because it was not THE D0 >>>standard format at the time (wouldn't want it to go away from underneath >>>us). A lot of the work that Yann has put in has been to fill out the >>>info.....he has added many classes for different L3 objects for example. >>>He says that most of what he has done he could translate to >>>TMBTree-style with just some tedious changes to how the classes are fed. >>>The ROOT macros should require very little modification. >>> >>>As long as Yann and Gordon stay in touch I think we can take good care >>>of the trigger content of the TMBTree. We will still complete the work >>>on trigsimcert for p17 so you should see an announcement of its >>>existence (+ features) soon. >>> >>>Thanks, >>>Dugan. >>> >>>On Mon, 2004-04-05 at 18:42, Gordon Watts wrote: >>> >>> >>> >>> >>>>Hi Dugan, >>>> Good point. Thanks. If we don't get in touch with you soon enough >>>>pester me again. I don't know how this is going to work out, but I'll >>>>try to make sure that we add it to the list of things. >>>> >>>> Cheers, >>>> Gordon. >>>> >>>>-----Original Message----- >>>>From: Dugan ONeil [mailto:dugan_oneil@sfu.ca] >>>>Sent: Monday, April 05, 2004 11:34 AM >>>>To: greenlee@fnal.gov; Serban PROTOPOPESCU; Gordon Watts; Yaroslav >>>>Kulik; deliot@hep.saclay.cea.fr >>>>Cc: Yann Coadou >>>>Subject: Trigger info and the data format working group >>>> >>>>Hi Herb, Serban, Gordon, Slava, Frederic >>>> >>>>I am very glad you have agreed to tackle the issue of a common ROOT >>>>format for D0! This is certainly needed. >>>> >>>>I wanted to make you aware that a couple of SFU people (led by Yann >>>>Coadou) have been writing tools to look at trigger output in the TMB. A >>>>standard tree is produced using some well-thought-out classes and some >>>>nice macros are provided to look at the contents (including every >>>>trigger subsystem, no more need to supply a toollist, reference >>>>distributions with chi-squared fits, etc.). While you are considering >>>>what trigger info needs to go into the new standard format I hope you >>>>will contact Yann about using (or adapting) his classes. It is better >>>>that we get the chance for input early in the process then rewrite >>>>everything from scratch. >>>> >>>>Do I understand correctly that this might be a wholesale reworking >>>>of the TMBTree (ie. it is still TMBTree in name only) or is it meant to >>>>be a small perturbation to that format? >>>> >>>>Thanks, >>>>Dugan. >>>> >>>> >>>> >>>>