I agree.  I would like to see a primary steering maintained for the current working version of the recon rather than new ones being made with copy and paste for every pass.  If we want to recover a version that was used then we can get it from SVN in the tag of the trunk made by the hps Java release.

I understand the desire to create these very specific steering files but the possible consequence is that the old ones get abandoned and eventually stop working as the code changes.

> On Oct 8, 2015, at 6:02 PM, Maurik Holtrop <[log in to unmask]> wrote:
> 
> Hi Jeremy,
> 
> I have noticed the same issue with some of the older steering files. It seems that when the software is updated, a new steering file is created for the new software, instead of updating the existing steering file. So instead of always running EngineeringRun2015FullRecon.lcsim, we have been creating new versions: *_pass1, *_pass2, etc. This does have the benefit that it is easier to later find which steering file was used for a pass, although it should be that you could get this as well by checking out the correct release. 
> 
> I would suggest that we always have “EngineeringRun2015FullRecon.lsim” as the latest greatest (and working) steering file. So right now we could set it equal to "EngineeringRun2015FullRecon_Pass2.lsim”.  And similar for the best MC steering file. That way we should always know which was the “best” steering file to run at any one point in the SVN history, even in between releases. Going forward, we should have a steering file for each data set if that data set requires changes from the default, and those should be maintained to work with every release.
> 
> Cheers,
>    Maurik
> 
> 
>> On Oct 8, 2015, at 7:41 PM, McCormick, Jeremy I. <[log in to unmask]> wrote:
>> 
>> Hi, Nathan.
>> 
>> The sandbox area is here.
>> 
>> svn://svn.freehep.org/hps/java/sandbox
>> 
>> And I'd make an "old_steering" dir under it for sandboxing the broken xml files.
>> 
>> Broken recon steering files I found today which I plan to move there include:
>> 
>> EngineeringRun2015FullRecon.lcsim
>> EngineeringRun2015FullReconOutsideIn.lcsim
>> HPS2014OfflineNoPileupRecon.lcsim
>> SimpleMCRecon.lcsim
>> 
>> All with the null pointer bug due to being out of date with the current reconstruction.
>> 
>> It seems to me maybe we should fix EngineeringRun2015FullRecon.lcsim though rather than sandbox it.
>> 
>> The MC recon files that seemed to work or at least run without crashing included:
>> 
>> EngineeringRun2014EcalReconMC.lcsim
>> EngineeringRun2015FullReconMC_Pass2.lcsim
>> EngineeringRun2015FullReconMC.lcsim
>> HPS2014OfflineTruthRecon.lcsim
>> 
>> And this one works fine that runs on SLIC output without readout sim:
>> 
>> /org/hps/steering/readout/HPSReconNoReadout.lcsim
>> 
>> And these ones seem to work okay that run on the data:
>> 
>> EngineeringRun2014EcalRecon_Pass1.lcsim
>> EngineeringRun2015FullRecon_Pass2.lcsim
>> EngineeringRun2015HitRecon.lcsim
>> GBLReco.lcsim
>> HPSTrackingDefaultsRecon.lcsim
>> 
>> I'm planning on adding to the build a few tests that will run the working ones on a few events so that we will know if they break.
>> 
>> I haven't checked steering files from the other different directories, but I assume that many are out of date and broken.
>> 
>> The user steering areas are probably fine as is and don't need to be checked.  It is each person's responsibility to keep those up to date if they wish.
>> 
>> If anyone knows of other steering files that are broken or should be checked during the Hudson builds please let me know.
>> 
>> --Jeremy
>> 
>> -----Original Message-----
>> From: Nathan Baltzell [mailto:[log in to unmask]] 
>> Sent: Thursday, October 08, 2015 4:31 PM
>> To: McCormick, Jeremy I.
>> Subject: Re: broken steering files
>> 
>> I vote for freely moving all broken steering files in "production" to a sandbox in svn and NOT deleted.
>> But first it needs to be announced where this sandbox is.
>> 
>> -nathan
>> 
>> 
>> 
>>> On Oct 8, 2015, at 4:06 PM, "McCormick, Jeremy I." <[log in to unmask]> wrote:
>>> 
>>> Hi,
>>> 
>>> I noticed that a couple of the old recon prod steering files don't work anymore.
>>> 
>>> EngineeringRun2015FullRecon.lcsim
>>> EngineeringRun2015FullReconOutsideIn.lcsim
>>> 
>>> with error
>>> 
>>> java.lang.NullPointerException
>>>      at org.hps.recon.utils.TrackClusterMatcher.isMatch(TrackClusterMatcher.java:207)
>>>      at org.hps.recon.particle.ReconParticleDriver.makeReconstructedParticles(ReconParticleDriver.java:264)
>>>      at org.hps.recon.particle.ReconParticleDriver.process(ReconParticleDriver.java:422)
>>>      at org.hps.recon.particle.HpsReconParticleDriver.process(HpsReconParticleDriver.java:132)
>>> 
>>> These are not backward compatible with the current version of the recon.
>>> 
>>> The Pass1 steering files seem fine.
>>> 
>>> Should the broken files be deleted or sandboxed?
>>> 
>>> I was planning to make an "old_steering" dir in the sandbox and move broken stuff there until it is fixed.
>>> 
>>> I'm going to try and test the MC ones next....
>>> 
>>> --Jeremy
>>> 
>>> ########################################################################
>>> Use REPLY-ALL to reply to list
>>> 
>>> To unsubscribe from the HPS-SOFTWARE list, click the following link:
>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1
>> 
>> ########################################################################
>> Use REPLY-ALL to reply to list
>> 
>> To unsubscribe from the HPS-SOFTWARE list, click the following link:
>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1
> 

########################################################################
Use REPLY-ALL to reply to list

To unsubscribe from the HPS-SOFTWARE list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1