Date:         Mon, 8 Apr 2002 09:41:10 -0400
Reply-To:     Joe McCrary <joe.mccrary@GADATA.ORG>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Joe McCrary <joe.mccrary@GADATA.ORG>
Subject:      Re: Are You a SAS Expert Test?
Comments: To: "James, Steve" <spj1@CDC.GOV>
Content-Type: text/plain; charset="iso-8859-1"

A few that I can think of:

1. Improperly sorted data (or unsorted data).
2. Different variable types (one numeric, one character).
3. Variables that are not being used to identify cases having the same name in both datasets, causing one to be overwritten (when you do not want that to happen).

-----Original Message-----
From: James, Steve [mailto:spj1@CDC.GOV]
Sent: Monday, April 08, 2002 9:38 AM
To: SAS-L@LISTSERV.UGA.EDU
Subject: Are You a SAS Expert Test?


Dear SAS-L
I recently saw a job advertisement for a "SAS Expert."  One way that they
were requiring that you demonstrate your mastery of SAS is  "Within SAS
Base, you must be able to name 25 different ways that the datastep can go
wrong in a merge statement. " Not that I'm interested in the job, but found
their question intriguing.  I couldn't come up with anywhere near 25 ways.
I'm wondering if the collective wisdom of SAS-L can meet the requirement of
this organization or if I'm just not as smart as I think I am.
Unfortunately I don't know their counting criteria.  I don't know if you
include syntax errors.  And if you count macro variables that resolve to one
of the problems below, then you've essentially doubled your list.  But I
thought it be good fodder for a pre-SUGI week discussion within the group.

My List
1. No BY Statement
2. Missing semicolon: e.g.  DATA ONE MERGE TWO THREE ; (Version 8 provides
some help with this)
3. Insufficient space to store the resulting data set
        a. this can either be WORK or permanent storage.
        b. on MVS, this can either be not enough space allocated or unable
to allocate any extents
4. Not handling the case when an obs from dataset A doesn't match an obs
from dataset B and vice-versa.
5. Many-to-many merges
6. Data set index is deleted after dataset is recreated by a datastep.
7. No obs in one or more of the datasets.
8. Merging more than two datasets is problematic (I think)
9. OBS= option on dataset or system option can prevent all observations from
being processed.
10. Mispelled variable name: either a BY variable or not.
11. One or both datasets not existing; mispelled dataset name ;

Steve James
sjames@cdc.gov