We should have a set of macros that implement different parts of nsIDocumentObserver. This will buy us two things: 1. Clean up the code by having a lot less empty functions that just return NS_OK 2. Makes it easier to change the signature of a function since we won't have to change so freakin' many function-implementations I'll try to do this in stages to make it easier to figure out how to group functions into the macros

I ended up implementing all macros and fix all implementors of nsIDocumentObserver. One interesting thing is that Begin/EndReflow is not implemented by any observers, and noone seems to call them either (only checked BeginReflow). So we should probably remove these notifications unless somebody has a greater plan that includes using them? Feel free to have oppinions on how i have grouped the functions into the different implementation-macros. I mostly grouped them according to functionality rather then analizying exactly how to get the minimum number of functions implemented outside the macros.

Those seem pretty reasonable to me...

Comment on attachment 107337 [details] [diff] [review] first stab sr=bzbarsky.

Comment on attachment 107337 [details] [diff] [review] first stab >Index: content/base/public/nsIDocumentObserver.h >+NS_IMETHODIMP _class::StyleSheetRemoved(nsIDocument* aDocument, \ >+ nsIStyleSheet* aStyleSheet) \ >+{ \ >+ return NS_OK; \ >+} \ >+NS_IMETHODIMP \ >+_class::StyleSheetDisabledStateChanged(nsIDocument* aDocument, \ Be consistent. You have the NS_IMETHODIMP on the same line as the class everywhere else; here they are on separate lines. But, actually I'd almost prefer you used this style (separate lines), since that is what content code uses everywhere else... On a separate note, I'm wondering if naming the macros |NS_IMPL_NSIDOCUMENTOBSERVER_IGNORE_*| is a better idea. It helps convey that we are really only implementing those methods just to ignore them because we don't care about them. NS_IMPL_NSIDOCUMENTOBSERVER_IGNORE_LOAD, NS_IMPL_NSIDOCUMENTOBSERVER_IGNORE_REFLOW, etc.... I suppose I will leave it up to you to decide, but I think those names are better (and I hope you have a good replace script if you agree :-) r=caillon

checked in. Thanks for the reviews