Tips and FAQ

The C++ compiler has to derived a unique name for all methods, and that should take into account namespace, the class name, possible template, the number and type of argument, the constness, etc.. This is called the C++ mangling of names. As a result, the symbol names as they appear in the binary file are difficult to decipher. You can use the c++filt utility to unmangle the name:

echo "_Z11OBUnmarshalR23SAM_SamDictionaryStructRPKhb" | c++filt
OBUnmarshal(SAM_SamDictionaryStruct&, unsigned char const*&, bool)
     

For KCC based releases, i.e. p14 and before, the demangler was called edg_decode.

There are two ways to setup the maxopt version of a release, but those two are not exactly equivalent :

• setup D0RunII xxxxx -O SRT_QUAL=maxopt
• setup D0RunII xxxxx; srt_setup SRT_QUAL=maxopt

The first formed is to be preferred, but the second form originally appeared in many tutorials. The difference is that the second form will end up using the RCP database for non maxopt (aka debug), which is fine as long as the the debug version of the release is present. This is rarely the case at CCIN2P3.

The RCP system has three different databases :

• The official database, available to everyone is constructed once a release is frozen. Still, if you're trying to read a file produced with a release that is more recent that the one you're using, the file may refer to RCP files unknown to your release. The problem can be alleviated by 'setup rcpfiledb'. Details can be found here.
• The build database, present while a release is not yet frozen. There is no solution to actually read data that was produced with an unfrozen release either with another release or when the release has been frozen.
• The personal database, that contains files from your local working directory.

In this example, 24484 is the id of the rcp file in database 1: 1 is the official database, 2 is the build database and 99 is the personal database.

• dsdump --notree FILE | wc -l
• evdump FILE | grep 'DSPACK data' | wc -l

This line splits the input file final.list into files tempoXXX.list that are 4 lines long.

awk 'BEGIN {count=0; filecount=0; x[0]=""}; {count=count+1;x[count]=$0; if (count == 4) {filecount=filecount+1;filename=sprintf("tempo%d.list",filecount); print filename; print x[1]"\n"x[2]"\n"x[3]"\n"x[4]>filename; print ""; count=0};}' final.list 

Another possibility is to use the split command line: this split the file input.list into sublist_aa, sublist_ab, sublist_ac, etc that are 4 lines long:

split -l 4 input.list sublist_