The International Organization for Standardization's OpenDocument Format standard is broken and needs to be mended, according to an expert who claims to have carried out tests on the format.
Alex Brown, a document-format expert who is convenor of the process to standardise Office Open XML (OOXML), posted a blog this week reporting the results of tests which he claims reveal that OpenOffice documents do not conform to the International Organization for Standardization's (ISO's) version of the OpenDocument Format (ODF) standard.
Speaking to ZDNet.co.uk this week, Brown, who reported similar problems between Microsoft Office 2007 documents and the OOXML standard, said the standardised version of ODF, known as ISO/IEC 26300:2006, "has a defect which prevents any document claiming validity from being actually valid. Consequently, there are no XML documents in existence which are valid to ISO ODF."
There is a critical flaw in the ODF schema defined by the Organization for the Advancement of Structured Information Standards (Oasis) and approved by ISO as ISO/IEC 26300:2006, according to Brown's blog, which means that no XML document can conform to the standard. Although the flaw invalidates ODF as a standard, it is relatively easy to fix, and Brown provided a defect report and suggested fix in his blog.
Even using a mended schema, Brown found in a "smoke test" that OpenOffice still does not produce valid standard documents: "This is to be expected and is a mirror case of what was found for Microsoft Office 2007." A smoke test is not a complete test, but the equivalent of starting up a car engine to see if it smokes, he explained.
Microsoft Office 2007 has not caught up with the ISO standard based on OOXML because changes were implemented in an ISO meeting, but OpenOffice has "bypassed" ISO/IEC 26300:2006, said Brown: "It aims at its consortium standard, just as Microsoft Office does".
Although OpenOffice is only one implementation of ODF, it is more popular than other ODF-based applications, such as KOffice, said Brown, and is therefore a good test. Brown took the same document that he used in his test of Microsoft Office 2007's conformance to the OOXML standard, saved it using OpenOffice and tested the resulting .odt (ODF format) file. It produced thousands of errors, most of which were very similar to each other.
The ODF community has yet to respond to Brown's findings and is currently working on a new version of ODF, version 1.2, for submission to ISO. Brown said his suggested change should be built into the ISO standard, based on ODF 1.0. "Technically this is not huge news," Brown told ZDNet.co.uk, "but ISO/IEC 26300:2006 is currently the only ODF standard and ISO should fix problems when they have been found."
Read this
UK standards body taken to court over OOXML
The UK Unix & Open Systems Users Group has launched a High Court challenge against the decision by the BSI to approve Microsoft's nascent document format...
Although Brown has been labelled as an OOXML supporter, he said he is in favour of all good standards and that, by offering a fix to ODF, he is actually supporting the document format. Brown referred to a post by Patrick Durusau, editor of the ODF standard, in which Durusau argued that standards supporters should promote and develop their own standards instead of "bashing" others.
"There may be flaws in ODF, but it is quite usable as it is," said John McCreesh, OpenOffice's marketing lead in the UK. He maintained that ODF is superior to OOXML, and that OpenOffice's team is uncovering problems with inputting OOXML documents: "If you do what the specification says, it doesn't look the same as it does in Office Word 2007. People want compatibility with Word 2007, not with some document that's alleged to say what the specification is."
Fundamentally, the issue is allowing user companies to be sure that their documents, which contain their intellectual property, will still be readable in 20 years' time, said McCreesh.
Talkback
According to engineers at OpenOffice, the problems Alex Brown raises are non-issues - brought about by errors in his tests.
Brown tested the ODF code using a validator without applying an option that disables unecessary checks, says OpenOffice. He also used a version of OpenOffice that saves to ODF 1.1 - a format which is not going to be submitted to ISO - they say.
In full, our response from OpenOffice runs:
"First, he called jing [an ODF validator] without the option "-i". If jing is called that way, it checks additional constraints defined by the 'RELAX NG DTD Compatibility' specification. ODF does not use this specification. jing therefore must be called with the switch "-i" to disable these checks. If called correctly, the document will validate.
"Second, he created the document using OpenOffice.org 2.4. OpenOffice.org 2.4 uses ODF 1.1. The error message about a soft-page-break is caused by the new element that has been introduced by ODF 1.1 to support conversions to DAISY Talking Book. We have already implemented this element at the request of the ODF Accessibility SC. If Alex had used the OASIS OpenDocument v1.1 schema, then this issue would not have
been reported. Please note however that we did not submit ODF 1.1 to ISO, because it is considered to be a minor update to ODF 1.0 only, and we were working already on ODF 1.2 at the time ODF 1.1 was approved."
Um, this 'OOXML expert' is the chair of the BSI committee that reversed the BSI's decision, isn't he?
I don't suppose this has anything to do with the action against the BSI in the High Court:
http://news.zdnet.co.uk/software/0,1000000121,39408917,00.htm
Whether it does, or is just yet another strange 'coincidence', surely he was aware of the flaws in his 'smoke test' as pointed out in Peter Judge's comment?
* FOOTNOTE *
Explanation of 'Chewbacca Defence':
http://en.wikipedia.org/wiki/Chewbacca_Defense
Rob Weir of IBM has criticised Brown's work in a <a href="http://www.robweir.com/blog/2008/05/odf-validation-for-dummies.html ">blog</a> post
http://www.robweir.com/blog/2008/05/odf-validation-for-dummies.html
Here's a summary:
1) Alex tried to validate an ODF using the wrong version of the schema. OpenOffice 2.4.0 writes out ODF 1.1, not ODF 1.0.
2) He also tried to apply validation rules that ODF does not claim to implement, specifically something called "Relax NG DTD Compatibility". If these checks are disabled, validation works fine.
3) The specific test case Alex used is invalid. If the document was already invalid OOXML, then why do we think that converting it to ODF will lead to anything other than an invalid document?
I guess this Brown chappie might be feeling a bit like certain SCO exec at the moment.
If OpenOffice people state that the version used by the Alex Brwon is the wrong version and he used the ISO/IEC 26300 version then obviously OpenOffice is not producing valid ISO/IEC standard files.
It is strange people now critisise the same principle as was apllied to the OOXML specification a week ago and when FOSS bloggers spread this everywhere as MS Office OOXML not being ISO/IEC compliant.
Obviously OpenOffice is also not ISO ODF compliant but at least for OOXML the standardization by ISO is still not completed and the ISO standardization of ISO ODF has been finished almost 2 years ago
Your 3rd point claims that the Office Open XML specifcation is not valid. This is not correct. The Office Open XML specification document is actually fully valid Office Open XML. It validates 100% correctly against the current XML schema provided with the Ecma-376 standard specification.
On point 2 a note that actually both Japan and Great Britain during the OOXML standardization submitted issues on the OOXML schema's allthough walid W3C schema's did not function with certain XML validators and OOXML was actually modified to accomodate some of those issues. For example issues jp-0002 and gb-0622.
Rob Weir never complained when counting those as issues on OOXML and that Ecma changed thing to accommodate those issues shows that in OOXML thing were changed to improve validation where Rob Weir just suggests diabling certain validations from the ODF schema's to make schemavalidators work. I guess he thinks validator issues are only issues when it applies to OOXML.
The ODF 1.0 version of the schema is the wrong version to use because the document itself clearly labels itself of being in the ODF 1.1 format. You can wish it to be an ODF 1.0 document all you want, but it doesn't make it so.
Remember, what Alex did with OOXML was different. He took a version N OOXML document and tried to validate it with a version N+1 version of the OOXML schema.
But what he did with ODF was take a version N+1 ODF document and tried to validate it with an version N of the ODF schema. This is an entirely different operation. One test is testing the backwards compatibility of the schema, the other is testing the backwards compatibility of document instances.
If you did the same test with ODF, taking the version N document (say the ODF 1.0 standard itself) and validated it with the version N+1 schema (ODF 1.1 in this case) then you would find that it works perfectly.
When I save a document in OOo, Lotus Symphony, or KOffice, I see no option to save the document as ISO/IEC:26300. I have only a single option per filetype.
The problem here is that the international standard is N but none of the full-featured tools available are writing to N; they will only write to N+1 which is not now and to my knowledge never will be an international standard. ODF 1.1 was never submitted to JTC 1 and plans are to submit ODF 1.2 without submitting ODF 1.1. But if history be any guide, before ODF 1.2 is finalized as an international standard, the full-featured ODF apps will be writing to ODF 1.3, without retaining the ability to write to ODF 1.0, 1.1, or 1.2.
This might not be such a problem if ODF 1.2 did not break compatibility with ODF 1.0 and 1.1 (the list enhancement "feature") or v. 1.2 specified a compatibility framework -- including preservation of relevant metadata -- for backward and forward compatibility among implementing apps. But watching the ODF TC mailing list, I've seen no moves in that direction.
But the bottom line seems to be that support for international standards is irrelevant to major implementing vendors, for both ODF and OOXML. The message from both sides of that particular fence seems to be that if conforming international standard support is identified as a procurement specification, there will be no bidders, i.e., if they specify N support, they can't get it.
And even that analysis blinks past a host of other interoperability issues in both ODF and OOXML. Both camps are oblivious to the market requirement of interoperability, even among implementations of their favored "standard."
This post has been removed by a moderator.
Microsoft has already in an open letter committed to supporting the upcoming ISO/IEC version (which is likely to be 1-1 with the next Ecma version soonish as well).
I think it is interesting that Sun/OOo and IBM are not going for the international ISO standard but to a version they refused to submit to ISO.
ISO/IEC jct1 sc34 is actually very unhappy with the independant maintenace of OASIS not bringing in versions of OpenDcoument in line with ISO/IEC versions.
This entire debate shows why there should be only one ISO standard. Even if Alex Brown is entirely genuine in his views, there is always the nagging thought that perhaps he has some commercial vested interest in promoting OOXML.
Alex Brown recently gained a place on the advisory committee of the British Library, an organisation that just happens to provide Adam Farquhar as chair of ECMA TC 45 for OOXML. The British Library and Microsoft are close collaborators. So independent operator with no particular vested interest? Hmmm.
It's wholly inappropriate for a person in Brown's position to start making emotive and what seem to be erroneous comments.
An "international standard" is defined, in ISO/IEC Guide 2:2004, definition 3.2.1.1, as "[A] standard that is adopted by an international standardizing/standards organization and made available to the public."
All OASIS Standards (or for that matter Ecma Standards) are international standards by this definition.
I guess the question to ask, is what incremental value would you ascribe to rubber stamping by JTC1? Was the approval of OOXML an indication of quality that is worth the effort attaining? I have my doubts. As it stands today, the title "OASIS Standard" is a far greater indication of quality than "ISO/IEC Standard". SC34 wishes that they could be as open, as transparent and as innovative as OASIS is.
Politicized issues make fertile ground for obfuscation and digressions. The issue of ODF adherence to ISO raised by Alex Brown is different in context and scope to that of OOXML. Comments suggests the following:
'ODF 1.1 is not compatible but ODF 1.0 is.' The issue is one of REMAINING compliant during updates. The culprit this time is Daisy Talking Book. ODF will now be watch more closely in this regard.
'OOXML is not ISO compatible. It has never been.' This issue is one of BECOMING compliant. OOXML has been highly contentious in this regard.
Statisticians apply 'slight of hand' equally well as magicians. If asked to apply his 'smoke test' to ODF 1.0 with the correct variables, Brown's response may tell whether his 'smoke test' was designed for objectivity or as a 'smoke screen'.
Standardization and compatibility are big issues for a reason. Don't be distracted by a mouse (called Daisy) when there is a 6000p elephant in the room. To compare these two is disingenuous.
One can quibble over the meaning of an "international standard," particularly within the context of the Agreement on Technical Barriers to Trade and the Agreement on Government Procurement.
But the fact remains that there are no full-featured editors that write to ISO/IEC:26300 OpenDocument. Numerous governments have adopted ISO/IEC:26300 OpenDocument as their national standards and procurement specifications and Microsoft has committed to conforming ISO/IEC:29500 Office Open XML support in MS Office.
Those facts set up the interesting possibility that Microsoft will have the only office suite that writes to an approved ISO/IEC standard and that the full-featured ODF editors remain legally ineligible for government procurement specifications that designate ISO/IEC:26300 as the standard that must be supported.
In other words, StarOffice, Lotus Symphony, OpenOffice.org and the other clones of the latter's code base are one legal challenge away from being blackballed legally from the government procurement process globally, allowing Microsoft to step into the breach in a few months, if it in fact carries through on its commitment to provide conformant support for ISO/IEC:29500.
To be sure, I do not know whether Microsoft would instigate such a legal challenge. But were I legal counsel for the companies that chose to support ODF v. 1.1 without preserving support for ISO/IEC:26300, I would make sure my clients were aware of just how legally vulnerable they are.
You claim OASIS is better that ISO/IEC but in your article you have stated that Alex Brown was wrong to use the official OASIS Committee Recommendation of RelaxNG DTD compatiblity in his validation method because it was not an ISO standard.
Why should OpenOffice then use ODF 1.1 which is also not an ISO standard ?
The evolution from odf 1.0 to odf 1.1 is simple and transparent. its entirely not the same thing as defining a new standard with thousands of pages of documentation.
Actually the current Ecma 376 standaard is already higly compliant to the upcoming ISO standard of Office Open XML.
In a previous validation test of an existing Ecma 376 document against the not yet released ISO transitional schema's, for a many thousands of pages long spec document only 84 validation errors were found.
So the upcoming ISO/IEC 29500 standard is already very similar to the current Ecma standard and document would only require very little change to be converted 1-1.
That certainly applies to ODF 1.0 and 1.1 as well but where OOXML is moving towards an ISO version ODF has actually moved away from an ISO version.
This post has been removed by a moderator.
The reason OOXML is undesirable is not to me a technical issue so getting bogged down in technically detailed arguments that most people don't understand is simple obfuscation. This seems to be the MS strategy. The fact is that single standards are better than multiple standards. Take power sockets. Within a country there is generally a standard but with international travel, adapters are now necessary for many people. There is little argument that having a single standard would be more convenient and less expensive for consumers. The argument about value in competition between standards in mature technologies is simple tosh. odf was established first and MS could easily have supported it. MS has more resources to do this than anyone else and there has been plenty of time. Obfuscation suits them because it buys them time maintaining their monopolistic hold on desktop office users for a bit longer. Unfortunately for them, the world is changing and their desktop dependent business model is looking shakier by the minute as more things migrate to the web. But that is their problem not a problem for the rest of us so no need to bail them out.
My objection to Relax NG DTD Compatibility, is not that it is an OASIS Committee Specifcation rather than an ISO Standard. My objection to apply that specification is that ODF 1.0 does not claim to conform to the part of that standard that was applied to Alex.
You can't just take a random standard and apply its constraints to ODF and say that ODF is broken because it does not conform to that other standard. For example, ODF 1.0 does not conform to XML 1.1, VRML, SGML, OpenFont, OOXML, or PDF/A. But we never said that we did.
ODF is defined by what the text of ODF standard states. It is that simple.