User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.10; rv:37.0) Gecko/20100101 Firefox/37.0 Build ID: 20150415140819 Steps to reproduce: I have an ordered list element which contains 3 list items. The ordered list is display: flex and the list items have no specified order, which causes them to show up as "0" instead of "1-3". <ol> <li>first</li> <li>second</li> <li>third</li> </ol> ol { display: flex; /* or inline also causes this error */ } ol li { flex: 1 0 auto; list-style: inside none decimal; } Actual results: List elements appear as 0. instead their correct number. Expected results: List elements should have incremented.

Workaround is to specify the count. ol { display: flex; /* or inline also causes this error */ counter-reset: level1; } ol li { flex: 1 0 auto; list-style: none; } ol li:before { content: counter(level1) ". "; counter-increment: level1; }

Chrome (Blink) & Opera 12.16 (Presto) give expected results (numbering the list entries as 1,2,3). Firefox Nightly numbers the list entries 0,0,0. Not sure why offhand.

Bug 4522 may be related, but in general the issue is that the list-numbering logic all lives in blockframe, no?

I'm not sure. We get it right if I make the <ol> "display:table", and we've got approximately the same block involvement with a table-<ol> vs. a flex-<ol>. So there's apparently some way to get this working with non-block <ol>. (Technically tables will give us a layer of blockification for the anonymous table-cell; I thought maybe that block was helping for tables. But if I add a similar block around each <li> or around all of the <li>'s and use display:flex, stuff's still broken.)

Ah, but as bz points out on IRC: in the "display:table" case, the anonymous block frame (for the cell contents) has its mContent pointer set to the <ol> content-node, so it does list numbering. Whereas, with "display:flex" and an explicit <div> around the <li> elements, that explicit <div> block does *not* have its mContent set to the <ol> content-node. (it's set to the <div>). So, no list numbering. So yeah, looks like a dupe of bug 4522.

Ah yes, and this was filed (for flexbox) a few months ago as bug 1129490. I thought it rang a bell, but I didn't search with the right terms to find that bug. :)