- From: Florian Rivoal via WBS Mailer <sysbot+wbs@w3.org>
- Date: Sat, 15 Jul 2017 17:12:02 +0000
- To: public-new-work@w3.org
The following answers have been successfully submitted to 'Call for Review: Scalable Vector Graphics (SVG) Working Group Charter' (Advisory Committee) for Vivliostyle Inc. by Florian Rivoal. The reviewer's organization suggests changes to this Charter, and only supports the proposal if the changes are adopted [Formal Objection]. Additional comments about the proposal: I strongly believe SVG is a key technology that needs to see continuing work, and would love nothing more than to support the charter for a healthy SVG working group. However, despite some small improvements since the previous proposal, I still think this charter is not good enough and is setting the group for failure. First, despite calls after the failure of previous charter for a broad discussion involving not only a handful of powerful companies, but the whole SVG community including invited experts, this charter seems to have been developed based on private discussions only. There have been no requests for comment to the www-svg list, or even to the narrower mailing lists for working group members (which would presumably still include all members as of the time the last WG charter expired). That does not necessarily makes the charter bad, but even a good charter won't do much if the group lacks attendees. To partially compensate for this lack of public discussion, As I believe input from non browser key contributors into this charter is essential, I reached out the Amelia Bellamy-Royds, invited expert, key contributor to SVG and co-editor of SVG2, while writing this. Her comments and suggestions significantly enriched my review, and as we ended up agreeing on mostly everything, this review also represents her views. Moving on to substantive issues about the charter: 1. Timeline: 1.1 An updated CR for SVG2 in August 2017 does not make much sense. There have been a handful of edits since the last CR that could be worth republishing. However, there are 138 open issues in github, including 95 tagged SVG-CORE (which means they affect the subject matter of the chapters in the main spec, although that may include feature requests and proposals in addition to issues). I find it hard to believe these are going to be triaged and resolved and get a disposition of comments within a month or so. 1.2. The timeline for getting SVG2 to PR is beyond optimistic, and well into impossible territory. SVG2 is a massive spec, it currently does not have a test suite. I invite people to check how long it took for CSS2.1, a specification of comparable complexity, to go from first CR to PR. Even if the spec is edited to only include features with 2+ implementations, there are countless details and edge cases within those features which need to be tested to demonstrate true inter-operability. And if the spec is going to be edited to remove features, then the August 2017 date for a revised CR is even more unrealistic. 1.3. The timeline for SVG-AAM with a CR by February 2018 (and PR by June) is not a whole lot more credible. Currently, there are some major spec issues waiting on feedback, and very little implementer interest. Widely inaccurate timelines and milestones are nothing new, and a fair few charters have that issue. So long as they don't affect the ability of the WG to work, it is regrettable that we put garbage there (especially if we are later going to evaluate the success of the WG against that timeline), but it's not in itself a blocking issue. However, in the case of this specific charter, it interacts very poorly with the question of scope (see below), turning it into grounds for objection. 2. Scope: The charter says: > As a primary focus in this charter period, the group will concentrate on the > stabilisation and interoperability testing of the core SVG 2 specification. > > "As a secondary focus, the group may address modules for new graphical > features for SVG, only once SVG 2 is at the Proposed Recommendation stage" > and once there is broad consensus on adding each such feature to the Web > Platform. A requirements document will be used to collect together these > features. I am 100% in support of non SVG2 work being a secondary focus. Working on that at a lower priority than SVG2 seems the best thing to do. However, this phrasing forbids working on it at all until SVG2 hits PR, which means that in the meanwhile, there will be no acceptable place to have any discussion on new features. This risks splitting the SVG community between maintainers and innovators, if it doesn't kill the innovation part entirely. It may also mean that the WG might be reluctant to split off from SVG2 a feature that would be slower than the rest to mature, since it could then be considered a thing you cannot work on at all until the rest is done. This in itself would be pretty bad even if that was for the short 1 year period the charter suggests, but given that I am convinced that taking SVG2 to PR will take several times as long. Here is a suggested alternative, that keeps finishing SVG2 as the top priority, without being quite as suppressive of forward looking work or as dismissive of people wishing to do that work. > As a primary focus in this charter period, the group will concentrate on the > stabilisation and interoperability testing of the core SVG 2 specification. > As part of that testing, features which do not meet the stability and > interoperability requirements for a Proposed Recommendation may be moved to > separate specification modules, work on which would remain in scope, but at a > lower priority. > As a secondary focus, the working group may compile feature requests and > proposals for new graphical features for SVG, and will provide a public forum > for discussing these ideas in order to be able to determine broad consensus > for adding such features to the Web Platform. However, spec development and > testing work on new feature proposals is not a priority until SVG 2 is at the > Proposed Recommendation stage. 3. Testing 3.1. "test as you commit" This is a laudable practice. Following this practice is likely to have positive effects in the long on the quality of the specification and on iteroperability. However, test writing and test reviewing are chronically understaffed activities throughout W3C. Unless multiple people (a single person cannot review their own tests) are committing to spend significant time for a period of several years to write tests and to review other people's tests, this will significantly slow down work on the specification, possibly bringing it to a standstill. Note that my main worry is not so much that people will fail to write tests (although that may very well happen as well). Rather, it is that tests written by editors mail fail to get reviewed for prolonged periods of time, stalling progress and demotivating these editors. This is even more likely to happen to contributors who do not work for major browser vendors and cannot drag in co-workers to review their tests. This is a problem on its own, but this probable slowdown also makes the prohibition on non SVG2 work until SVG2 PR even worse. I suggest that this clause be removed from the charter, and that this be left to the chair’s discretion. I would encourage him to make the Working Group follow the "test as you commit" approach if and when practical, but making it mandatory is likely to cause problems. 3.2. Focus of the test suite The charter says that the SVG test suite should concentrate on changes relative to SVG 1.1 Second Edition. I do not believe this to be a good idea. Although the existing SVG 1.1 test suite will be a useful starting point, it is not anywhere near a comprehensive test suite for the un-changed parts of the spec. The key motivation for focusing on SVG2 above other work is to achieve greater interoperability. This can only be realized through thorough testing. Artificially lowering the bar so that we can hit TR milestones more easily is counter productive. The reviewer's organization intends to participate in these groups: - Scalable Vector Graphics (SVG) Working Group The reviewer's organization: - intends to review drafts as they are published and send comments. - intends to develop experimental implementations and send experience reports. - intends to develop products based on this work. - intends to apply this technology in our operations. - would be interested in participating in any press activity connected with this group. Answers to this questionnaire can be set and changed at https://www.w3.org/2002/09/wbs/33280/SVG-2017/ until 2017-07-17. Regards, The Automatic WBS Mailer
Received on Saturday, 15 July 2017 17:12:08 UTC