Post on 05-Jan-2016
transcript
inline XBRL: is it Evolution or just a Future
Herm Fischer
Mark V Systems LimitedApril 26, 2013
Topics
• What’s so hard about cranking up inline?– HMRC problem– EDInet (FSA Japan)– at SEC?
• What gets submitted to authority• What gets validated
What is inline XBRL
• Old fashioned html document– Decorated and organized by filer
• Fact values marked with tags– Numbers (with transformation/scaling)– Strings etc.
• Contextualizing information is hidden– Periods, Dimensions, Units, etc.
inline XBRL
Filing with inline XBRL
HMRC Use Case
Model of inline XBRL
HTML supported
• No Doctype in html (tricks for old IE versions)• Semi-custom html-4 (validated by xsd)– No html-5 stuff
SEC Filing
• Before inline XBRL– 8k/10k (1 html/text file)– Instance in XBRL (1 file)– Schema and linkbase files
• After inline XBRL– 8k/10k (1 html file(s) with inline markup)– Schema and linkbase files
Filer Manual Accommodations
• There will be duplicate facts– Most accurate likely prevails (?)
• May be other EFM issues
Validations
• HTML syntax (schema, not DTD)• Inline transformations and contextualizations• XML of extracted instance• XBRL 2.1, dimensions• Edgar Filer Manual• (Not by SEC: public pillory on the cloud)
What gets rendered to consumer
• The original source html– Has decorations, style, and ordering of filer
• The extracted XBRL instance– Normalized information and appearance– Normalized ordering of information– Uniform rendering like among accessions
• **both of these (my opinion)
Round trip re-rendering
• From inline source– Extract XBRL facts, contextualizing information– Render normalized way (e.g., SEC website)
include inline in normalized HTML rendering• Should be losslessly recyclable
like a Möbius strip• Never get original decorations
back
Changes for Japan
• EDInet will use inline XBRL “document sets”– Each schedule a separate html document– Footnotes and notes in separate document at end
• Needed extensions for– Date formats (emperor’s birthday-based dates)– Shared numbers (“same-as-left column”)– Fact-explanatory-fact footnote-links– Text continuation (strings split across sections)
Copyright © 2012 XBRL Japan Inc., All Rights Reserved
2. Abbreviated notations - Situation in Japan
15
A notation of “Same as left” is really often used in JP typical reporting documents, and we have similar problems as UK vendor reported.
Number of issued shares
At the end of fiscal year
(2012-01-31)
At reporting date
(2012-04-26)
fact=“Same as left”?fact=8176452
Number of consolidated subsidiaries
Prior consolidated fiscal year(2009-04-01~ 2010-03-31)
Current consolidated fiscal year(2010-04-01~ 2011-03-31)
fact=“Same as left”?fact=3
Practical issues in Japan
• Some consumers have XBRL-aware tools– Adequate inline support exists
• Told that FSA will not send out XBRL instances– Instance semantically identical to inline
• Independent software guys need support– Want open source only, no licensed software– Original prototype inline extractor not usable– So do they have to scrape from HTML? Can they?
What can we learn from usage?
• ? document set filings (> 1 html)• ? download extracted instances• ? available open source software• ? render from XBRL without decorations