sPHENIX discussion of software

[Thomas K Hemmick <tkhemmick AT gmail.com>]

Sorry to but in and possibly say something stupid, but in teh past I found that some rebuilds require users to call upon the nuclear option:

• Delete all locally compiled things in your build and install areas.
• build and install all again from scratch.

Basically the assumption is that when there is a deep enough change in the core libraries, make fails to catch all necessary rebuilds and even "make clean" is insufficient.  A completely empty install directory (libs and includes) and a completely fresh rebuild of all local code always did the trick for me.

Tom

On Mon, Apr 12, 2021 at 8:53 AM Huang, Jin <jhuang AT bnl.gov> wrote:

That is very interesting!

 

The JenkinsCI test used the default macro + the DST Saving flag on. And here is the list of nodes :

https://web.sdcc.bnl.gov/jenkins-sphenix/job/sPHENIX/job/test-DST-readback/3454/artifact/macros/detectors/sPHENIX/Fun4All_ReadBack.log/*view*/

including

      TRKR (PHCompositeNode)/

         TRKR_HITSET (IO,TrkrHitSetContainer)

         TRKR_HITTRUTHASSOC (IO,TrkrHitTruthAssoc)

         TRKR_CLUSTER (IO,TrkrClusterContainer)

         TRKR_CLUSTERHITASSOC (IO,TrkrClusterHitAssocv2)

 

One suggestion is to run on the master build only, not to run “source  /opt/sphenix/core/bin/setup_local.csh $MYINSTALL” to avoid local inconsistency.

 

Another caveat is that Jenkins run on newest master branch, instead of new build. But yesterday’s build is a pass too.

 

Cheers

 

Jin

 

 

______________________________

 

Jin HUANG

 

Physicist, Ph.D.

Brookhaven National Laboratory

Physics Department, Bldg 510 C

Upton, NY 11973-5000

 

Office: 631-344-5898

Cell:   757-604-9946

______________________________

 

From: Christof E Roland <cer AT mit.edu>
Sent: Monday, April 12, 2021 8:45 AM
To: Huang, Jin <jhuang AT bnl.gov>
Cc: Pinkenburg, Christopher <pinkenbu AT bnl.gov>; sphenix-software-l AT lists.bnl.gov
Subject: Re: [Sphenix-software-l] DST reading broken

 

Hi Jin, 

 

I tried it interactivly using the nightly build.

 

source /opt/sphenix/core/bin/sphenix_setup.csh

and 

source  /opt/sphenix/core/bin/setup_local.csh $MYINSTALL with MYINSTALL set to an empty directory.

 

The DST file writing and the readback macro executed 5 min from eachother. 

 

the DST readback test splits reading and writing at which level?

I tried to write hits and the read things back to clustrize the hits. This fails. 

 

generating hits and clustering in one go works ok. 

 

Cheers

 

   Christof 

 

On 12. Apr 2021, at 14:39, Huang, Jin <jhuang AT bnl.gov> wrote:

 

Hi Christof

 

DST generation-readback with the current master branch is tested daily on Jenkins CI. This morning’s test at 4AM EDT has passed: https://web.sdcc.bnl.gov/jenkins-sphenix/job/sPHENIX/job/test-DST-readback/3454/

 

Suggest double check the consistency of the lib and builds, as different new.[1234] builds would be picked up from day to day.

 

Cheers

 

Jin

 

 

______________________________

 

Jin HUANG

 

Physicist, Ph.D.

Brookhaven National Laboratory

Physics Department, Bldg 510 C

Upton, NY 11973-5000

 

Office: 631-344-5898

Cell:   757-604-9946

______________________________

 

From: sPHENIX-software-l <sphenix-software-l-bounces AT lists.bnl.gov> On Behalf Of Christof E Roland
Sent: Monday, April 12, 2021 8:34 AM
To: Pinkenburg, Christopher <pinkenbu AT bnl.gov>
Cc: sphenix-software-l AT lists.bnl.gov
Subject: Re: [Sphenix-software-l] DST reading broken

 

Hi Chris, 

 

this is not about reading old files.

 

I can't read back new files that I wrote this morning with the latest build.

 

Did anybody try this?

 

Cheers

 

  Christof 

 

On 12. Apr 2021, at 14:28, pinkenburg <pinkenburg AT bnl.gov> wrote:

 

Hi Christof,

that's because of the broken schema evolution which is now fixed in the core software. The fallout of any such fixes is typically that you cannot read back old files.

I am working on a two step process to convert old DSTs (to salvage our MDC DSTs) which you could apply to these once this is ready. Otherwise either creating new DSTs or linking against an old ana build (before the code change last week) is the only option

Sorry,
Chris

On 4/12/2021 5:22 AM, Christof E Roland wrote:

Hi Everybody, 

I tried to write some hit files and read them back for clustering using the latest coresoftware/macros 

repos and this consistently fails due to the versioning of the ClusterHitAssoc. See error messages below.

Could so,eoen have a look?

Thanks 

   Christof 

Warning in <TStreamerInfo::BuildCheck>: 

   The StreamerInfo of class TrkrClusterHitAssoc read from file /phenix/u/bogui/RunOct20/Jan21Const/SvtxJan21_Pu50_020_2_10_1125.root

   has the same version (=1) as the active class but a different checksum.

   You should update the version to ClassDef(TrkrClusterHitAssoc,2).

   Do not try to write objects with the current class definition,

   the files will not be readable.

Warning in <TStreamerInfo::CompareContent>: The following data member of

the in-memory layout version 1 of class 'TrkrClusterHitAssoc' is missing from 

the on-file layout version 1:

   multimap<unsigned long,unsigned int> m_map; //

Warning in <TClass::Init>: no dictionary for class MvtxHit is available

Warning in <TStreamerInfo::BuildCheck>: 

   The StreamerInfo of class TrkrHit read from file /phenix/u/bogui/RunOct20/Jan21Const/SvtxJan21_Pu50_020_2_10_1125.root

   has the same version (=1) as the active class but a different checksum.

   You should update the version to ClassDef(TrkrHit,2).

   Do not try to write objects with the current class definition,

   the files will not be readable.

Warning in <TStreamerInfo::CompareContent>: The following data member of

the in-memory layout version 1 of class 'TrkrHit' is missing from 

the on-file layout version 1:

   double m_edep; //

Warning in <TStreamerInfo::CompareContent>: The following data member of

the in-memory layout version 1 of class 'TrkrHit' is missing from 

the on-file layout version 1:

   unsigned int m_adc; //

Warning in <TClass::Init>: no dictionary for class InttHit is available

Warning in <TClass::Init>: no dictionary for class TpcHit is available

input_line_202:5:21: error: allocating an object of abstract class type 'TrkrClusterHitAssoc'

         *ret = new TrkrClusterHitAssoc;

                    ^

TrkrClusterHitAssoc_Dict dictionary payload:39:16: note: unimplemented pure virtual method 'identify' in 'TrkrClusterHitAssoc'

  virtual void identify(std::ostream &os = std::cout) const = 0;

               ^

TrkrClusterHitAssoc_Dict dictionary payload:37:16: note: unimplemented pure virtual method 'Reset' in 'TrkrClusterHitAssoc'

  virtual void Reset() = 0;

               ^

TrkrClusterHitAssoc_Dict dictionary payload:46:16: note: unimplemented pure virtual method 'addAssoc' in 'TrkrClusterHitAssoc'

  virtual void addAssoc(TrkrDefs::cluskey ckey, unsigned int hidx) = 0;

               ^

TrkrClusterHitAssoc_Dict dictionary payload:49:6: note: unimplemented pure virtual method 'getClusterSet' in 'TrkrClusterHitAssoc'

    *getClusterSet(unsigned int layer, unsigned int phi_segment, unsigned int z_segment) = 0;

     ^

TrkrClusterHitAssoc_Dict dictionary payload:58:5: note: unimplemented pure virtual method 'getHits' in 'TrkrClusterHitAssoc'

    getHits(TrkrDefs::cluskey ckey) = 0;

    ^

TrkrClusterHitAssoc_Dict dictionary payload:60:24: note: unimplemented pure virtual method 'size' in 'TrkrClusterHitAssoc'

  virtual unsigned int size(void) const = 0;

                       ^

input_line_202:8:21: error: allocating an object of abstract class type 'TrkrClusterHitAssoc'

         *ret = new TrkrClusterHitAssoc[nary];

                    ^

input_line_202:13:29: error: allocating an object of abstract class type 'TrkrClusterHitAssoc'

         *ret = new (arena) TrkrClusterHitAssoc;

                            ^

input_line_202:16:29: error: allocating an object of abstract class type 'TrkrClusterHitAssoc'

         *ret = new (arena) TrkrClusterHitAssoc[nary];

                            ^

Error in <TClingCallFunc::make_ctor_wrapper>: Failed to compile

  ==== SOURCE BEGIN ====

__attribute__((used)) extern "C" void __ctor_0(void** ret, void* arena, unsigned long nary)

{

   if (!arena) {

      if (!nary) {

         *ret = new TrkrClusterHitAssoc;

      }

      else {

         *ret = new TrkrClusterHitAssoc[nary];

      }

   }

   else {

      if (!nary) {

         *ret = new (arena) TrkrClusterHitAssoc;

      }

      else {

         *ret = new (arena) TrkrClusterHitAssoc[nary];

      }

   }

}

  ==== SOURCE END ====

Error in <TClingCallFunc::ExecDefaultConstructor>: Called with no wrapper, not implemented!

_______________________________________________

sPHENIX-software-l mailing list

sPHENIX-software-l AT lists.bnl.gov

https://lists.bnl.gov/mailman/listinfo/sphenix-software-l

-- 

*************************************************************

Christopher H. Pinkenburg       ;    pinkenburg AT bnl.gov

                         ;    http://www.phenix.bnl.gov/~pinkenbu

Brookhaven National Laboratory  ;    phone: (631) 344-5692

Physics Department Bldg 510 C   ;    fax:   (631) 344-3253

Upton, NY 11973-5000

*************************************************************

_______________________________________________
sPHENIX-software-l mailing list
sPHENIX-software-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/sphenix-software-l

 

_______________________________________________
sPHENIX-software-l mailing list
sPHENIX-software-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/sphenix-software-l