I'm entering this as a bug so that the code doesn't get lost, since I've pulled it out of my tree. I still think we should consider doing something like this in the future, so I'm going to move this to MFuture. The code I'm going to attach is a few functions (already slightly out of date) that do mode selection based on strict parsing of the DOCTYPE. They go to noquirks for unknown DTD's from 3 sources (W3C, ISO, IETF), but quirks for unknown ones from other sources. What is done with a given set of DTDs could easily be changed. The parsing is done as close to the SGML spec as I could get, although there are a few things I don't handle fully.

wont fix; pulling strictDTD support.

Reopening. We still use doctype parsing for layout quirks.

Marking [DOCTYPE] for easy searching.

Nominating for mozilla 0.9. I'd like to get this in (revised, due to bug 55264) for mozilla 1.0, at it should have significant testing beforehand. The current code is bad because it's just wrong. We should be testing equality of doctypes, not searching the DOCTYPE's public ID for certain strings such as numbers that we think are versions or searching the system ID for certain strings that are currently used as the file naming conventions for storage of DTDs in W3C specs (e.g., "strict.dtd"). While it can be demonstrated to be correct on many common examples, we've already seen one example of serious failure (bug 55916), and there will probably be more.

For lack of comments from rickg, assigning to myself.

I've been trying to fix this in a sensible way involving only parsing the DOCTYPE in one place. However, the htmlparser design makes this very difficult.

Removing patch keyword based on dbaron's comments that the current patch is no longer valid. Good luck with getting this to work :-) Gerv

updated qa contact.

Reality check. Moving out to 0.9.1.

David: Do you still want to do something with this?

*** Bug 82261 has been marked as a duplicate of this bug. ***

I would still like to do this, but 0.9.2 should be a low-risk milestone, so moving to 0.9.3. I guess we'll just ship 0.9.2 with the mess we currently have, unless someone feels strongly that I should try to do this for 0.9.2.

what's the current status?

Patch is incorporated into the patch on bug 55264.

Fix checked in 2001-09-08 11:37 PDT (see bug 55264).