- From: Peter Moulder <peter.moulder@monash.edu>
- Date: Sun, 16 Dec 2012 13:48:42 +1100
- To: www-style@w3.org
On Thu, Dec 13, 2012 at 12:22:42AM +0000, Sylvain Galineau wrote: > But why should we use - or would want to use - the UA stylesheet to > work around an initial value that is incompatible with existing > content? That's the part I don't follow given how regularly we choose > initial values (or define an initial auto value) to ensure compat with > existing content. What's so specific about these properties that this > should not occur here? It would help if you made explicit referene to fantasai's existing answer to this question, namely (paraphrased) that different UAs have different backwards-compatibility concerns, and some would be better off with an initial value of 2. Print-based UAs in particular are likely better off with 2. I wonder whether we can find an initial value that tends to do the right thing, something that tends to give two lines when those lines are "normal text" and it doesn't result in greatly under-full pages/fragments, and tending to give only one line if that line consists of huge inline blocks or replaced content, or when two lines would cause a greatly under-full page. An initial value like '2em' would get much of that, and might reduce compatibility concerns enough that it could be the initial value for both browser and print-based UAs. I hesitate to address the under-full concern yet, but one possibility would be for widows/orphans/break-* values to allow an 'unless-underfull(30%)' modifier. A weakness is that that doesn't distinguish between making one or five pages 30% under-full (e.g. if a UA looks five pages ahead to balance under-fullness). Something else relevant to adding under-fullness as a consideration for page breaks is that UAs are already at least allowed to infer *additional* places where breaking should be avoided if this can be done without making a page (or fragment) too under-full. That possibility is enough to allow improved pagination, but of course doesn't by itself provide either user control or consistency between UAs. pjrm.
Received on Sunday, 16 December 2012 02:49:08 UTC