On Thu, Oct 09, 2008 at 01:27:22PM +1100, David Gibson wrote: > On Tue, Oct 07, 2008 at 11:50:38AM -0500, Scott Wood wrote: > > On Tue, Oct 07, 2008 at 12:41:03PM +1100, David Gibson wrote: > > > > Instead of /addnode/, how about an alternate version of (or option to) > > > > /merge/ that merges the second tree with the contents of the first, > > > > > > Um.. I don't entirely see how this variant of /merge/ would differ > > > from /addnode/. > > > > /addnode/ can only add one node at a time, and has an different syntax > > than normal for representing the node to be added (name is split from > > body). /mergeunder/ could add any number of nodes and/or properties, > > and would use more normal syntax. > > Um.. in that case I don't see how this differs from the original > /merge/.
I guess the issue is that /merge/ is incompletely specified. The example showed it taking anonymous trees as parameters; are the parameters always wrapped as such, or can named nodes be passed directly? In the latter case, /mergeunder/ was intended as a way of dereferencing the former argument so as to place things inside it rather than create a new anonymous node containing both, without needing to know the name/unit-address of the named node. An alternative would be a function to extract the name of non-anonymous nodes, which could be used to build a matching subnode in the second /merge/ argument. > > We probably should have added comma-delimiting when we switched to > > decimal-by-default. > > Maybe, but its done now. I really don't think requiring parens aroud > expressions is that big an imposition. Agreed; I may find it as ugly as you find identifier prefixes, but we don't have much of a choice if we want to retain compatibility. -Scott _______________________________________________ devicetree-discuss mailing list [email protected] https://ozlabs.org/mailman/listinfo/devicetree-discuss
