From lechner@cs.uml.edu  Sat Oct 28 16:24:13 2006
From: Bob Lechner <lechner@cs.uml.edu>
Message-Id: <200610282045.k9SKjQ0Z211476@saturn.cs.uml.edu>
Subject: Re: [UML_CS522_f2006_OOA] Re: Schema file for asgnt3 [candidate key names?]
To: UML_CS522_f2006_OOA@yahoogroups.com
Date: Sat, 28 Oct 2006 16:45:26 -0400 (EDT)
Cc: 06f522

RJLRef: $PH/06f522/asgnt3/CKspecsInputFdbk_ns_rjl061028.txt

> From: "Nitin Sonawane" <nitin.sonawane@gmail.com>
> X-Yahoo-Profile: nitinsonawane
> Sender: UML_CS522_f2006_OOA@yahoogroups.com
> Subject: [UML_CS522_f2006_OOA] Re: Schema file for asgnt3
> Reply-To: UML_CS522_f2006_OOA@yahoogroups.com
> Hello again,
> 
> I just realised that Prof. Lechner already has schema files setup in his
> metatest3 directory called metaCKplusSISrev061014.sch . Please ignore the
> file I emailed and use this one instead (plus my file also has a few syntax
> errors in it).
> 
> Thanks,
> Nitin.

Thanks Nitin. I'd like to remind you that test data is available
in SISrev061011.dat, as a start for the DB to be queried against.
I modified this from SIS test data but it may also contain
some errors and needs to be inspected and/or pr_loaded/dumped
and compared for errors and/or surprises.


Re: your CK specs below, the unit test specs should declare 
(in a comment or post-condition) what keyExists(query) should 
return as a value, and what its clients shold expect.
(Clients may want to insert new or find old records by a query CQ.)

Multiple levels of test coverage can be formally defined.
Boundary testing is quite useful: do tests suceed and/or fail 
wheen they are supposed to? (I.e. when input queries intentionally
contain values that SHOULD cause true or false return from 
keyExists(thatQuery). Does the pre-loaded SIS DB affect the test
more than it should (with 0, 1, or >1 duplicates, for example)?

Another point: 
In real life, CKspecs may come from some requirements analyst
in a format beyond your control. THat doesn't prevent you from
writing a filter that makes its use more convenient.
awk (or perl, or sed, or vi) can edit your CKspec  
format below (which has no CKname?) into a simple space-separated
list of words, which has a trivial parser (already generated
as pr_parse(), tailored to ANY .sch table type in pr_load.c).

You can declare such a word list in a CK table format; 
If you call the field names fname1, fname2, etc. (up to some max) 
in the schema you can access them in pr-parse statements 
like atoi or pr_set_str or encode  which follow  
the word separation statement hcg_parse(buffer, hcg_t, &idx);
See pr_load.c:pr_parse() in $PH/06f522/asgnt3/lechner/metatest3.

Exactly the same technique can extract field values from a 
similarly declared and pr_parsed row of a CQ query instance.
However, you'll need a table type for each signature variant 
to get max mileage from generated code. Again, see how pr_parse
uses the value lists that it separates (atoi, pr_set-ster, encode).

Bob Lechner

> 
> 
> On 10/28/06, Nitin Sonawane <nitin.sonawane@gmail.com> wrote:
> >
> > Hello folks,
> >
> > Please find attached the schema file for assignment 3. This file is
> > essentially a concatenation of metaschemaCK.sch and SISrev061011.sch.
> > Following  Prof. Lechner's convention, I've added some explanatory comments
> > at the top. I've also verified that this file is properly parsed by chgen.
> >
> > Also, following up on Prof. Lechner's comments, here are the candidate
> > keys
> >
> > Department DE
> >     (deptNo)
> >     (DEname)
> >     (DEhead)
> >
> > Student AA
> >     (studentId)
> >
> > Term TM
> >     (TMname)
> >
> > Faculty FY
> >     (DeptId, FacyName)
> >
> > Course CO
> >     (DeptId, CourseNo)
> >
> > Address AD
> >     (AAid)
> >
> > StudentTerm RT
> >     (AAid, TMid)
> >
> > CourseTermSection CT
> >     (FYid, TMid, COid, sectionNo)
> >
> > Prerequisite PQ
> >     (COid1, COid2)
> >
> > CourseSchedule CS
> >     (SDbldg, SDroom, SDtime, SDdays)
> >
> > Enrollment ET
> >     (RTid, CSid)
> >
> > Will follow up with a preliminary version of the test driver soon.
> >
> > Thanks,
> > Nitin.
> >