1. CEN-CENELEC Workshop Agreements (CWA) on XBRL.
(Katrin) New deliverable in development, currently as German draft, about:
Metamodel transformation (mapping) between SDMX 2.1 and XBRL.
(Ignacio) The collaborative space on www.wikixbrl.info (working area) as well as www.cen.eurofiling.info (deliverables, project documents, source code and so on) has been updated with the formal deliverables. The only remaining update is:
2. EBA news. None.
3. EIOPA news. None
Next Eurofiling Workshop / XBRL EU Day possibly in the week of 5 May, Rome. TBC.
Large Instance Processing (LIP).
(Katrin) Bundesbank contacted with large reporting entities asking for the expected size of XBRL instance documents. In the case of FINREP, it has been observed about 9,000 rows to report, correponding to a size of 300 Mbytes. The tests has been performed up to 100 Mbytes. Sizes up to 1 Gbyte are not to be discarded. Lack of performance is especially critical on instance validation (reception is not an issue).
The performance of 32 bits servers is questioned, but upgrade to 64 bits is a mayor task.
Interchange of experiences is welcomed.
(Joel) XBRL community is working on streaming/serialization specifications.
(Goto) Bank of Indonesia is a reference case for large instance processing. However, the case is not fully comparable due to the use of tuples instad of dimensions. Formulae processing is a hot topic.
(Martin) Performance is very different for different XBRL architectures. Assertions have a large impact. Usual test cases with “fake data” (all “0”, or all “7777777777”) are not very realistic, as a huge number of errors arise in assertions. More realistic test cases would be more than welcomed.
(TO BE CONTINUED…)
Andreas Weller (EBA)
Ignacio Boixo (Eurofiling coordinator)
Ignacio Santos (Bank of Spain, ES)
Joel Vicente (CoreFiling, UK)
Katrin Heinze (Eurofiling coordinator)
Martin DeVille (UBPartner)
Masatomo Goto (Fujitsu, UK)