PHD

[MattO1@aol.com]

In a message dated 11/19/98 1:32:30 PM Central Standard Time,
Robert.Edis@creatcomp.com writes:

<< 
 I have just experienced a very good reason for having PHDmaint around.
 My morning has been spent rebuilding a Y2K compliant version of a 80,000
 line dictionary using PDL.  Due to the size and the inability of
 Power2000 to handle large files, the PDL extract had been split into 11
 files, 1 for system & usages, 1 for security, 4 for elements and 5 for
 file/record definitions.  I was on the 4th and largest file definition
 PDL went I got an error.  Two numeric items had mistakenly been declared
 with a date usage in the 2nd element PDL file.  I had spent about 4
 hours getting to this point in the PDL compile process.
 
 Now what do I do?  In the 'new' order of things I would have to delete
 the dictionary so far created and start all over again!  Four hours
 wasted.  Instead I used PHDmaint to quickly go into the new PHD, find
 the offending element definitions and change the datatype back to
 numeric, zoned size 6.  In all less than five minutes work.
 
 Now let's see, 5 minutes vs 240 minutes.  I think I know which both I
 and the people who pay my billing rate believe is better.
 
 Please Cognos, put PHDmaint or something as good or better in version
 8.2n of PowerHouse.  Please, please.
 
 Blue
  >>
Couple of comments.  First you seem to have made a very good case for having a
source code version of the dictionary.  I don't know how many changes were
required in you 80,000 line dictionary to make it Y2K compliant, but I'm
assuming quite a few.  Several hundred at least.  Potentially several
thousand.  

The prospect of making that number of changes on a completely interactive
dictionary is frightening.  That is the primary reason (as I understand it)
that PDL was eventually included with the VAX.  PhD  just was not practical
for mass changes of metadata.  There are other very good reasons as well, but
that one is a killer.

Now, as far as breaking the PDL code into several pieces.  Even if you did not
have to do it for Power2000, it's still not a bad idea.  I've seen many sites
do that long before there was a Power2000.  Regardless if that is a good thing
or not, I think you are mistaken on one point.  If the 4th file has an error,
that does not mean you have to throw away all your compile time.

With PDL you do not have to start from scratch every time you compile.  If you
notice, the CREATE DICTIONARY statement creates the dictionary, but it is
empty until you compile some PDL and issue a LOAD statement.  Then the PDL
that has been compiled clean UP TO THAT POINT can be loaded.  In other words,
if you place a CREATE DICTIONARY statement in the front of your 1st file, and
a LOAD at the end of each of the 5 or 6 files, you can perform incremental
loads of the dictionary.  If there are any errors, then the LOAD will not
execute, but that will only effect the last file you attempted to compile.

Hopefully, that makes sense.

As for me, I love PhD.  It's great for making quickie changes, but that makes
it very easy to screw up something too.  Like Allison Hamilton mentioned
earlier, I never change the production dictionary "live". I always make a
copy, make the changes to the copy, and then after testing, copy the new
version over the old.

PhD is a great convenience.  PDL is a necessity.  I could live without PhD.  I
would be lost without PDL when it came to mass changes. (Which seems to be
what I'm doing a lot of lately.)

Just my 2 cents worth. 8-)
Matt Ohmes      
Matt.Ohmes@Cognos.Com;  MattO1@AOL.COM  (that's 'Oh1' not 'Zero1')

Cognos Corp.
909 E. Las Colinas Blvd,  Suite 1900
Irving, TX  75039 
(972) 556-1122    Fax (972) 556-1004

"Matt is a fool and his views not only don't represent those of Cognos or any
of it's employees, but furthermore we refuse to admit he even works for the
company, let alone that he knows anything relevant.  Don't trust him and don't
lend him money!"
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Subscribe: "subscribe powerh-l" in message body to majordomo@lists.swau.edu
Unsubscribe: "unsubscribe powerh-l" in message to majordomo@lists.swau.edu
powerh-l@lists.swau.edu is gatewayed one-way to bit.listserv.powerh-l
This list is closed, thus to post to the list, you must be a subscriber.