Created attachment 284787 [details] Patch (WIP) Yet another step toward bug 156537. This covers: - lspace - rspace - minsize - maxsize
Created attachment 284854 [details] Patch (applies after bugs 160241, 160245, 160239 and 160190)
Comment on attachment 284854 [details] Patch (applies after bugs 160241, 160245, 160239 and 160190) This pattern of having a "dirty" flag in various values is not a good one. Because it means we have an incorrect value that we could read and we have to be careful not to write code that looks at it without looking at the dirty flag first. A much better pattern is to use Optional and set to null when the value needs to be recomputed. This ties together the dirty flag with the value it guards in a more explicit way.
(In reply to comment #2) > This pattern of having a "dirty" flag in various values is not a good one. > Because it means we have an incorrect value that we could read and we have > to be careful not to write code that looks at it without looking at the > dirty flag first. A much better pattern is to use Optional and set to null > when the value needs to be recomputed. This ties together the dirty flag > with the value it guards in a more explicit way. OK, I've opened bug 160400 and also updated the first patch in the series (bug 160239) to use Optional<T>. So let's consider these patches first and I'll rebase all the patches afterwards.
Created attachment 285104 [details] Patch (applies after bug 160190)
Created attachment 285122 [details] Patch
Committed r204037: <http://trac.webkit.org/changeset/204037>