Mozilla does not support cursor: auto. For example, HTML {cursor: default} INPUT, A, TEXTAREA {cursor: text}, which I use to specify the platform default cursor on all but the user input enabled elements won't work. See the specified URL for detailed a testcase.

This isn't a lyout issue, so changing component to Style System

This will depend on bug 1916, "9 Values for the Cursor Property (CSS2) do not work", M15 -- until it is fixed, there will be no way to properly test cursor: auto. See the comment by dbaron@fas.harvard.edu on 1999-10-17 for why.

It doesn't depend on 1916 at all. Cursor: auto only requires that those cursors that are used by default are supported. These cursors are: 1. pointer 2. default 3. text All of these are supported so the dependency is wrong.

webmaster@richinstyle.com: That is an incredibly complex test page. Please create a simple test case which is clear, and attach it to this bug report. Thanks.

Copying comment from Pierre entered into bug 51113 (relevant to this bug): It seems to me that nsTextFrame::GetCursor() should iterate through all the parents (not just the first one) to find the first one which doesn't have NS_STYLE_CURSOR_AUTO (and if we don't find any, set NS_STYLE_CURSOR_TEXT), a mechanism a little bit similar to what we have for 'cursor-select' in nsFrame::IsSelectable(). When it's done, could you make sure that bug 30971 (cursor: auto does not work) and bug 20080 (cursor: inheritance bug: not inherited inside floated span) are still valid? Thanks.

Block moved M18 bugs to mozilla0.9.1

Netscape's standard compliance QA team reorganised itself once again, so taking remaining non-tables style bugs. Sorry about the spam. I tried to get this done directly at the database level, but apparently that is "not easy because of the shadow db", "plus it screws up the audit trail", so no can do...

I had it all wrong in my comments above - and reading the code, I wasn't the only one. Cursor is an inherited property: unlike visibility, we don't need to look at the parents' style. I'll attach a fix and a testcase. Marc, Waterson: please review.

OK, I see that the text frame is setting the auto cursor to TEXT, but I don't quite understand why the base frame is setting it to DEFAULT. Does that mean that links won't get the hand, or does that get set elsewhere? I'd have to apply the patch and test it, but in this case I trust that you did just that, so [s]r=attinasi.

Sure, I tested it. "cursor:auto" doesn't mean "inherit your parent's cursor" but "set the cursor to whatever is appropriate for you". For a generic frame, the default cursor is, well, 'default' (the arrow). A very few frames override GetCursor(). It's because they set it in the css files, like html.css for the links.

Moving to m0.9.2

fixed

VERIFIED using Pierre's testcase.

I think this fix caused bug 88688 per Marc's fear: "Does that mean that links won't get the hand, or does that get set elsewhere?" Can someone verify and re-open this bug or comment in the other bug that this is not the case?

IU backed out Pierre's changes locally and it did not fix the problem with cursor:auto on a link, so I don't think this fix actually regressed it, rather it never worked at all...

Reopening: cursor:auto does not work on a link (anchor). See testcase in bug 88688

Milestone reality check.

Marked fixed again. Bug 88688 is not a regression but a different instance of this bug that was not covered by my patch.