- From: Simetrical <simetrical@gmail.com>
- Date: Tue, 5 Aug 2008 11:43:33 -0400
- To: "Jens Meiert" <jens@meiert.com>
- Cc: www-style@w3.org
On Tue, Aug 5, 2008 at 3:36 AM, Jens Meiert <jens@meiert.com> wrote: > At least this is more maintainable than e.g. "Conditional Comments" … > However, "user agent sniffing" means a maintenance problem in its own, > and history already shows that it never works reliably [1]. If individual sites misuse a CSS-based UA-sniffing capability, negatively affected browsers will provide a way to work around that, by spoofing. That's no different from the situation today: minority browsers like Opera still need to (and do) provide a robust spoofing mechanism to deal with poorly designed sites. They would just have to (eventually) extend that existing capability to spoof CSS conditional statements as well. The point is that UA sniffing is inevitably the cleanest way to do things in some cases, and in those cases a clean and robust way to do it should be provided. That it will be misused is no argument. Similar capabilities are already misused, but providing more ways for poorly-designed sites to misuse them will make them no more poorly designed. On the other hand, it *will* make the well-designed sites simpler to author and maintain, and that's the point here. > Also, I wonder if these approaches are not standards-adverse in principle. No, they're not. Any robust (programming) standard needs to acknowledge that there will be some deviations from the standard, and provide ways to handle them. This isn't standards-adverse, it's just realistic. CSS is already better than most in its provision for vendor-specific prefixes, but it could be improved significantly if it took the simple step of adding rendering engine/platform-sniffing capabilities. It would probably be a few extra paragraphs in the specification.
Received on Tuesday, 5 August 2008 15:44:10 UTC