- From: marbux <marbux@gmail.com>
- Date: Sun, 25 Nov 2012 06:11:57 -0800
- To: "public-markdown@w3.org List" <public-markdown@w3.org>
On Sat, Nov 24, 2012 at 5:09 PM, marbux <marbux@gmail.com> wrote: >> Then after almost a week I don't understand what is a "profile". > > I'll take a swing at answering that, since it's a definition I > researched about 3 years ago. > > Given a specification A, a profile is a subset (B) of the > supersetting A specification. In effect, a sub-specification to which > implementers can claim conformance. A superset specification can have > multiple subsetting profiles, optimally layered so that each profile > incorporates by reference all profiles with a smaller feature set in a > linear fashion until the full specification is reached. Hence the need > to begin by identifying and specifying a "core" profile and working > our way outward to a full specification. And by logical extension, > once the full specification is supersetted, it becomes a profile of > the new superset specification. > > Profiles are essential to interoperability. Assume for a moment that > we were dealing with the OpenDocument Formats ("ODF") rather than > Markdown. Google Docs and Zoho Writer do not support the full feature > set supported by OpenOffice.org. They are lightweight editors. But > there is no lightweight editor profile of ODF. Hence Docs and Writer > can send documents to be processed by OOo without fear of markup loss > but documents created with OOo cannot be processed by Docs and Writer > without concern for data loss. > > But consider the difference if ODF had a lightweight editor profile > and a conformance requirement that a conformant implementation of a > superset profile specification must process subset profile content as > if it were the superset profile content. Then if Docs and Writer > conformed to the lightweight editor profile, documents could be > round-tripped between them and OOo without fear of markup loss. > > And were OOo equipped with the means to select which profile the > document is to be written to and by selecting a mode that makes > features unavailable not supported by the lightweight editor profile, > OOo users could originate documents to be shared with Docs and Writer > without concern for markup loss. For example, in OOo open a new HTML > document and notice that the GUI changes to make unavailable features > not supported by HTML. The same could be done for new lightweight > editor profile documents. > > So in my view, the goal of defining a core profile is to identify the > minimum feature set to which all Markdown implementations must conform > to claim conformance to that profile. That does not, however, rule out > supporting more features that are defined in an intermediate profile > or are application-defined. It simply means that a conformant > implementation must be capable of processing Markdown documents as > defined by the core profile, unless we add some sort of metadata to > indicate that a document conforms to the core profile, which seems to > be a non-starter given the Markdown goal of legibility as a plain text > document. I should have mentioned a couple of things. 1. Despite a thorough search, I found no authoritative definition of a "profile" in the IT specification sense. The explanation I gave is based on snippets that seemed reliable, conversations with experts, and what I could infer from specifications that have been profiled, notably, the W3C Compound Document Working Group's WICD profiles and the conformance section of its Compound Document by Reference Framework specification. <http://www.w3.org/2004/CDF/>. 2. Experts disagree on whether a full specification is also a profile. The majority I discussed this with insist that a full specification is not a profile, that a profile is necessarily a sub-set. This area of disagreement seems to be entirely nit-picking; i.e., it does not affect the requirements of a specification or profile in any way. In my explanation, I rode with the majority but I could have as easily taken the opposite view without any effect on the profiles or a full specification.
Received on Sunday, 25 November 2012 14:12:25 UTC