Can I derive the libname from a Data Source or Data Source Item prompt output?

[jakarman]

Ok Scott,
You described the question you are trying to solve very detailed and at the high level.

I can also see more of the differences we are thinking and the experiences we have.

Your target-group customers are using AMO and needing a query on fact/dimansion tables to the detailed records.
This is giving a lot of options on that level:

- Information map studio.

  Defnine the joins as new views. This will work whenthe kind of views are stable. Normally the case with cubes/Olap.

- When you have installed the successor of ODBC that is OLDB that is the interface to cubes in Excel to SAS.

  This is separate cleint at the SAS installation.
- SAS(R) 9.4 Statements: Reference  You could use a datastep view.

  This is a datastep as you are already using but the view is like a dataset but will be delayed to run until you access it.

  The idea is the same as information map studio.

Insising on dynamic prompts with SP-processes you have to split up the perfromace question.
1/ how long does it take for the drop-downs appear

2/ how long does the prosessing take.  (no comment for that I assume)

The proc delete reference is mentioning the performance disadvantage SAS  did with proc datasets and even more with the SAShelp.v approach.

They are first trying to get all information off all involved datasets into memory (a lot of io) thereaftere a selection is made by you.
This will happen every time you are using that, as nothing is cached.
For performance there is Always a trade off in IO memory and cpu. The goal is minimize IO wihtout stressing the other two.

So solve the IO missing cache to program that yourself. Make a copy of all information on regular schemes.

As you are using cubes that should no big challenge as building them is a far more heavy job to do.
The effect of the WS server is just checking and informing as a second time the WS should already be running.

[jakarman]

scott, Now your hi-jacked subject to security.

The best place to secure assets is at the assets themselfes. This is an open door sorry for that.
As most assets areon the OS-level, txt html rtf files are also assets the OS-security is mandatory.

OLAP/Cubes is failing to define the assets in a clear way. That is by design of cubes.
The cubes contain information you may and may not have access to. At the same time the higher aggreagation levels are allowed, this is a contradiction.

This contradiction trying to solve is a terrible challenge as it implies row-level security.

Some RDBMS systems have the option of this granularity. The problem is that rowl-level security is dependent of content/value.

SAS has row-level security as a programmatical approach implemented. By that the original table must be protected to having no access by them in an other way.
SAS introduced metadata-bound libraries. trying to force metadatacontrols.  Stlli when the original table can be copied it could be decrypted.

It is bad seurity as based on not knowing how it works and very small limited user base, so  bad guys are not interested. Same applies for the pwencode promotion. As SAS is trying to fulfill programmer requests without thorougly thinking and design they are easily doing a bad security job.