Hi, Peter:
I don't have a crystal ball or a direct line to the developers, but I hope not. PROC MEANS/SUMMARY is good for "quick in/out make me the data, get me the report, I don't need it in a particular structure". I see PROC MEANS/SUMMARY as being the first step in the road to other reporting procedures.
I look at it like this: I can put a single egg (data) in my Cuisanart food processor (TABULATE) and whisk it that way, or I could just use my simple handheld whisk or egg beater that gets the job done too (MEANS/SUMMARY).
When you get to the point with PROC MEANS that you say, "Oh, this is wonderful, but I want the values of WHIRLYGIGS nested inside the values for GEWGAWS and I want a new page for every unique value of FRAMISTAT. And could I have the MIN, MEAN and MAX statistics underneath a column for each unique value of KNICKKNACKS. And I need a count of TOTAL WHIRLYGIGS by KNICKKNACKS as the final column." That's when I haul out the Cuisanart.
PROC MEANS has the TYPES and WAYS statements, which I love, because they simplify the way that you can get different combinations of crossings of the CLASS variables -- in a more shorthand format -- than having multiple TABLE statements with TABULATE. And, MEANS/SUMMARY has the ID statement to get other vars into the output dataset. Now that TABULATE can create an output dataset, I think that's totally cool and I have used the feature -- not as much as I just use PROC MEANS for OUTPUT datasets or even PROC REPORT for OUTPUT datasets. I guess my opinion is that PROC MEANS/SUMMARY combination made sense, since by the time they were rewritten to use/create output objects, they were essentially the same procedure.
I like having PROC MEANS and PROC TABULATE as 2 separate procedures. They are enough different that I can see MEANS appealing to one audience for its simplicity of use and TABULATE appealing to another audience for its power and elegant TABLE statement syntax. I know that TABULATE can potentially have a large number of crossings for the CLASS variable (2^31-1) and I'm not sure on whether MEANS has the same limit or not, but that might be one area of difference.
So, no hard and fast answers, just my .02...
cynthia