[css-variables] RE: CSS Variables in Last Call

It's been months since I considered writing this email. I was hoping that, over time, my concerns would appease and that it somehow would make sense, at some point. The sad truth is that it never happened. Worse, some of the decisions I considered wrong did lead to other issues that just feel even worse (to name the most recent one: the request made to the TC39 committee to have randomly-ordered hash maps to cope up with the "allow utf8, don’t mess with utf8" thing) and confirmed me in my beliefs.

We have been running for so long now a faux-last-call with a draft full of issues that I don't even see the point. The sad reality is that those issues do not even cover a majority of the issues being currently discussed on the list. And for what it is worth, I don't even believe it (the set of the issues currently being discussed) to be a good representation of the things that should be discussed. 

This specification does not serve best its official intent, with preprocessors being much more powerful and efficient when it comes to handling the static cases the spec says it aims to solve, by providing the right tools and evaluation mechanism and not just a simple memory storage. Either way, it doesn't solve well the other use cases that were envisioned: web components custom properties, and polyfills; and I'm very doubtful the current approach will ever solve them properly anytime soon. One does not easily build a city on a slippery sponge. 

I just asked myself the only question that really matters: knowing how much I supported the idea of author-defined properties, and knowing that a failure at this point may take months - at least - to recover, would I (if I was given the responsibility, which hopefully isn't the case) object to this specification? The answer is: yes, I would.

Do whatever you want with this, this is not an official last call comment or whatever, this is just the (personal) feelings of someone who really care about this and in all honesty believes the spec is headed the wrong way because it totally fails to see any big picture pattern and define its most basic concepts in reusable terms. 

I wonder how many people actually cared about this spec, and how much it was the result of a thoughtful discussion process as opposed to a vague approval of a plan that seem to lead to the solution of one of the oldest longstanding css issues (eventually followed by reports for the found implementation issues); but I won't take a wild guess.

Tab, please don't hate me :-/

± -----Message d'origine-----
± De : Philippe Le Hegaret [mailto:plh@w3.org]
± Envoyé : mercredi 24 juillet 2013 12:28
± À : www-style@w3.org
± Objet : CSS Variables in Last Call
± It appears that I forgot to send the Last Call announcement for this draft...
± CSS Custom Properties for Cascading Variables Module Level 1
± http://www.w3.org/TR/2013/WD-css-variables-1-20130620/
± All persons are encouraged to review this document and send comments to
± the www-style mailing list (you're on it if you're reading this). When sending
± e-mail, please put the text “css-variables” in the subject, preferably like this:
± “[css-variables] … summary of comment…”
± Comments would be appreciated by September 4, 2013. (ignore the July 18
± date on the draft).
± This module introduces cascading variables as a new primitive value type that
± is accepted by all CSS properties, and custom properties for defining them.
± CSS is a language for describing the rendering of structured documents (such
± as HTML and XML) on screen, on paper, in speech, etc.
± Thank you,
± Philippe

Received on Monday, 2 September 2013 04:56:22 UTC