[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [office] Re: Draft I-7-13 Review of *:name attributes
Patrick, I completely agree with the concordance value of having all of the attributes and elements cross-coupled this way. However, having a technical tool to support this and having the document be organized around it are two different decisions. I am also concerned that this approach makes the ODF specification too much scheme-tied and it may lead to conceptual difficulties of a different kind as we attempt to sort things out. I agree that the current organization has made certain difficulties apparent. With regard to the current rush to issue 1.2 I believe this may be a detriment to our quick publication and a distraction to those we hope will volunteer competent review. It is challenging, for example, to know exactly what has changed between 1.1 and 1.2 (not what is new), and whether those changes are inadvertent (especially the breaking changes). In software development, there is always a trade-off between refactoring a system, the risk of destabilizing the system, and the risk of new problems. I think that is a valuable lesson. I do not propose to change the approach. I do think it is going to be challenging to act on what it shows us and to find and repair the regressions that seem to have occured in some instances. I very much like the cross-referencing that is built into the document and pray that we will be able to leave that linking in the delivered specification. - Dennis PS: I also agree that context needs to be handled implicitly. I did not mean to suggest tacit understanding when speaking of context. We need lots more explicit context. PPS: So long as draw:name is used as the attribute for affixing a name to an entity and for occurrences of references to an entity so named, we've already got a mess. That draw:name is also used to affix names in different families of entities raises interesting questions about the level at which the affixed names must be unique. We also have situations where the same affixed name is referenced via an attribute that has a different name without explicitly knowing which (kind of) entity is being referred to. [The XSLT can't help with that either, unless we had an interesting way to decorate the schema with RDF so the identification of and reference to entities is modeled.] This will come up again, I suspect. -----Original Message----- From: Patrick Durusau [mailto:patrick@durusau.net] Sent: Monday, January 05, 2009 15:57 To: dennis.hamilton@acm.org Cc: Rob Weir; Michael Brauer; ODF TC List Subject: [office] Re: Draft I-7-13 Review of *:name attributes Dennis, I think you fail to appreciate the difficulties with the prior organization. [ ... ] So, by listing the attributes in this manner, it calls attention to differing definitions of the same attributes. Perhaps in many cases not something we can change in this version but it does make those differences apparent. [ ... ] As far as "context," I think that is a very dangerous thing to rely on in standards. Take the chart attributes for example. Some of them are meaningful if and only if they are used with other attributes. But, we never said that in ODF 1.0 or ODF 1.1, but relied upon "context." That is the sort of "context" that needs to be rooted out of a standard that is meant for others to rely upon and not only those who authored it. So, yes, there is a lot of repetition but that is because one or more of the definitions for a particular attribute are different. As definitions are reconciled, the amount of repetition can be reduced. Hope you are having a great day! Patrick PS: The excellent cross-referencing and automated checking of the elements and attributes are due to Michael's XSLT skills. Dennis E. Hamilton wrote: http://lists.oasis-open.org/archives/office/200901/msg00026.html [. .. ] > PS: It occurs to me that if one wanted to expedite completion and reviews of ODF 1.2 quickly, the breaking into three parts was a good idea but the entire restructuring of the treatment of attributes has been very costly. Informative and useful in one way, but very costly in other ways. (The cross-referencing is great.) > > Dennis E. Hamilton [ ... ]
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]