Re: The specification development process

There are two questions here: one about tools and publication workflow, and one about project management in the sense of defining requirements, scheduling work, criteria for completion etc.

The tool chain for building the specs is undoubtedly complex, which creates a barrier to entry; on the other hand every feature in the toolchain is there because it served a useful purpose at the time and by-and-large, it still serves that purpose today. The tooling has a learning curve but it greatly helps our productivity and the quality of our deliverables. There have been a number of attempts over the years to write helpful documentation describing the publication workflow, and some of those documents are present in the repository; inevitably, they are out of date. Most people seem to find when using tools like these that the best approach is copy-and-paste, trial-and-error, and asking questions, rather than reading documentation. It's encouraging to me that 8 or 9 people have successfully raised PRs. It's also encouraging that we have dropped some of the complexity that was present in the W3C days: we no longer generate applets from the grammar, we no longer generate namespace documents, etc. The document editor is also less of a bottleneck since under Git the editing process is more open to participation.

As for process, Norm rightly says that it's very hard to treat standards making as a conventional project. W3C did try to scope activities by writing a requirements document, and that can have some value because it gives a WG a way of deeming some of the "left-field" proposals (pardon my baseball terminology) as being out of scope; but in the end, the WG and its activity is driven by its members. The WG considers the proposals that its members put on the table. We will get to a point where we have to do some kind of "code freeze", start rejecting new ideas and putting them in a "v.Next" bucket, and start closing off unfinished business. We haven't reached that point yet; though we should perhaps put a bit more effort into marking some of the issues that have been raised as "rejected" or "won't fix".

Michael Kay
Saxonica

Received on Tuesday, 6 June 2023 16:40:09 UTC