Zebra enhancement ideas
       ________________________

          R.Raja 11-Aug-95

1)Protecting against Unprotected links in Zebra
____________________________________________________

One of the main objections to the Zebra  is the occurrence of unprotected links.
This is a self-inflicted wound in D0. It occurs mainly in muon and tracking
codes. The calorimeter code uses the following system to protect itself against 
unprotected links.

It puts all permanent links in its code in a common block called /ZLINKC/.
This common block is declared to Zebra as a link area by a call to MZLINK.
Any routine that uses a link, has this common block included in it.
Instead of a declaration such as

          INTEGER LPELC

          LPELC = GZPELC()

The code would have

          INCLUDE D0$INC:ZLINKC.INC


          LPELC = GZPELC()

The INTEGER LPELC declaration is contained in the common block /ZLINKC/.
This scheme can be imposed on tracking and muon code by inventing similar common
blocks

/ZLINKM/ and /ZLINKT/ with all the links declared. We would then write a
scanning program that would go through all the code looking for occurrences of
unprotected links and replacing them with the appropriate INCLUDE statement.
This can also be done manually if preferred.

An abbreviated version of /ZLINKC/ is given below.
C----------------------------------------------------------------------
C-
C-   Created   2-MAY-1989   Rajendran Raja
C-
C----------------------------------------------------------------------
      INTEGER LNKMX
      PARAMETER(LNKM
X = 100 )          ! maximum number of links
      INTEGER CSTLNK                    ! Structural links
      INTEGER CRFLNK                    ! Reference links
      COMMON /ZLINKC/CSTLNK(LNKMX),CRFLNK(LNKMX)
C
C the user can add more links  by equivalencing to CSTLNK,CRFLNK
C arrays.
C
C
       INTEGER LCAD1,LCAD2              ! Structural links
C...
C
       INTEGER LMTRH,LCAPH
       INTEGER LPARH,LCACL
       INTEGER LCACH,LJETS
       INTEGER LJTSH,LJPTS
       INTEGER LPELC,LPPHO
       INTEGER LPNUT,LPVES
       INTEGER LBANK                    ! generic bank address
C
       INTEGER LCAHTR,LCAPHR                   ! Reference links
       EQUIVALENCE (CSTLNK(1),LCAD1),(CSTLNK(2),LCAD2)
C
C...
C
       EQUIVALENCE (CSTLNK(3),LHITS),(CSTLNK(4),LCAHT)
       EQUIVALENCE (CSTLNK(21),LPELC),(CSTLNK(22),LPPHO)
       EQUIVALENCE (CSTLNK(25),LPJET),(CSTLNK(26),LCMAP)
       EQUIVALENCE (CSTLNK(28),LFAKE),(CSTLNK(29),LGEAN)
       EQUIVALENCE (CSTLNK(31),LHMTE),(CSTLNK(32),LHMTP) 
       EQUIVALENCE (CSTLNK(35),LPNUT1),(CSTLNK(36),LPNUT2)
       EQUIVALENCE (CSTLNK(38),LJNEP)                    
       EQUIVALENCE (CSTLNK(41),LZFIT)
C
       EQUIVALENCE (CRFLNK(1),LCAHTR),(CRFLNK(2),LCAPHR) !REF LINKS

----------------------------------------------------------------------------
2)Enhancement of .ZEB files

Our ZEB files can be modified to contain more information.
A clear improvement would be to add Mnemonic names for variables as well ranges
of variables if needed. These mnemonics can then be used to build Fortran
structures, NTUPLES or ADAMO entities from Zebra banks using a program.
Another possibility is to convert the ZEB files to the DZDOC format.

3)Zebra Checker- Very often, the ZEB files are out of date and do not correspond
to the banks in length or format. A program can easily be written to match the
MZFORM info in the bank , with the ZEB file info.


4)DBANK should be enhanced to be platform independent. A version to be made
available for Level 3 nodes as well.

5)ZBANK be enhanced to use the ZEB file information for formatting the print
routines it writes.

With the above modifications, most of the short comings of zebra can be
circumvented. While it may not be as elegant as Object oriented schemes, in the
long run, it is my view that the first priority of D0 is to write the upgrade
tracking software and then to retrofit the rest of D0 software to the new
scheme. I do not believe we have the skilled manpower needed to accomplish the
huge task of essentially re-writing RECO and do the upgrade tracking software
and improve all the ID algorithms as well in the proposed time frame.