• stations and chambers
• detection elements
  • sectors
  • slats
  • planes
• ddls
• buspatches
• manus
• pads

Stations (st) and Chambers (ch)

5 stations, 2 chambers each

= 10 chambers

10 chambers

Detection Elements (de)

Elementary parts of the detector

(re-moveable and hence should be aligned)

156 units

Sectors and slats

Stations 1 and 2 : sectors (a.k.a. quadrants)

8 units, 2 types

Stations 3,4 and 5 : slats

148 units, 19 types

Planes

Each DE has two planes

Bending (b) & Non-Bending (nb) plane

A plane is a set of pads

Motifs

A group of <= 64 pads

One FEE card

283 types (171 for slats)

work

DDLs & Buspatches

GBTs and CRUs

Voltages

Numbers recap (Run 2)

• 20 DDLs
• 156 DEs
• 888 buspatches
• 16828 manus
• 1064008 pads

Numbers recap (Run 3)

• ~25 CRUs
• 156 DEs
• ~600 elinks
• 16828 dual sampas
• 1064008 pads

Finding pads

• by (ix,iy)
• by (x,y)
• by (fee,channel)

a.k.a. Segmentation

Linking R/O and segmentation

• DE <-> DDLs
• DE <-> buspatches
• DDL <-> buspatches
• buspatches <-> FEE

Linking DCS and segmentation

• manu <-> HV
• DE <-> LV

Each role used in different parts of the code

e.g.

• find by position : clustering
• R/O to pad : raw data decoding
• HV/LV : data filtering

Text files

• 666 (!) text files
• several different formats (custom)
• interpreted w/ custom readers
• used to construct objects hierarchy

OCDB

• Text files [ in 1 ROOT obj in OCDB ]
• Still treated as text files

All or nothing

one single (big) set of objects

• in memory (but designed to be compact)
• all the time
• for all purposes

Data source format

• custom and small text files were nice
  • when developping the mapping
  • but mapping is now known and fixed
• still, human readable stays usefull

Use JSON as the only text format

• still human readable / writeable
• but parser(s)'s coming "for free" this time
• in whatever language

Limit the number of files

one per "detector level" (ch,de,seg,etc...)

Generators

• don't (necessarily) use JSON directly
• generate code for trivial/small things
• generate binary format(s) (e.g. FlatBuffers)

and

• be more "functional" and less OO ?

Split mapping

natural level of work(*) is DE

one (binary) mapping per DE

(*) up to tracking

Status : txt

• berg
• bp
• ch
• ddl
• de
• motif (for slats)
• pcb
• slat
• sector

... modulo tests, of course ...

see https://github.com/mrrtf/alo/blob/master/jsonmap/mch-mapping-convert-to-json.cxx

Next Steps

• code the reverse conversion txt
• to check all information is actually still there...
• (re)write tests
• generate some code and some binary format