> Hello again all IMAGE users,
>
This is another in the continuing series: 1998 SIGIMAGE
> Enhancement Ballot results discussion & call for comments:
>
The "Top 20" results are still available at:
http://www.csillc.com/sigimage/results.html
While giving 3000-L readers another week or so to continue
to respond to the first message in this series on 98-04-16
(on QUERY Item # 98I29), we'll get started on the next one:
The voter's favorite in both '97 and '98: Ballot Item # 98I01:
> ========================================
Bundled 16/32-bit ODBC direct to TurboIMAGE, without
having to go through Allbase.
> ========================================
>
> History of this item on the last two SIGIMAGE Ballots
(as per previous, ballot results prior to 1998 have been
"adjusted" to a 50-votes per ballot basis to match '98):
> 112 ballots total in '98; 74 ballots in '97; 65 ballots in '96:
>
> 1996: Not on ballot in '96 (bundled 32-bit ODBC driver
was, and that has been delivered: ODBCLink/SE).
> 1997: Rank = # 1; # of votes = 752; # of ballots = 44
> 1998: Rank = # 1; # of votes = 518; # of ballots = 45
>
> In 1998 one person gave this item all 50 votes; one gave
it 40; four gave it 25; six gave it 20; tapered off from there.
Perspective: *Not* counting the other "ODBC direct" item
(to MPE & KSAM files), in '97 Item # 98I01 was the winner
over the next closest item by a ratio of 3.5:1. In '98 it beat
out the next closest item (of any flavor) by a ratio of 1.5:1.
> HP item STATUS at time of the 1998 Ballot: For over a
year, everyone in CSY from GM Harry Sterling on down has
said they are not going to do this, because the solution is
available now from two third-party vendors. They plan to
concentrate on internal TurboIMAGE enhancements that
only HP can do.
But users keep voting for it, so it will remain on the ballot.
A LITTLE MORE ANALYSIS, leading up to some questions:
For the first time in '98 the SIGIMAGE Ballot allowed voters
to indicate the primary and secondary reasons why they voted
for a particular enhancement. There were five "reasons" you
could pick: Performance, Functionality, Ease of Use, Ease
of Maintenance, and Other. "Primary Reason" results for Item
# 98I01 were:
Functionality: 31
Performance: 06
Ease of Use: 05
Ease of Maint: 01
Other: 02
"Ease of Use" was the most popular "Secondary Reason"
picked for this item, by a clear margin.
SOME ADDITIONAL BACKGROUND (I have reason for this):
There were eight items in the "IMAGESQL Utility" section of
the Ballot in '98 (Item #'s 16-23). Individually, none of them
came close to making the "Top 20" enhancements this year
(cutoff for the "Top 20" was 94 votes)... HOWEVER:
*Collectively* the IMAGESQL Utility group did pretty well:
274 votes total.... So if "fix IMAGESQL" would have been
one enhancement request it would have ranked # 4 in 1998.
Anyway, short item description followed by # of votes:
Bundled retain/reuse attach parameters if DETACH: 58
Improve functions for managing authorization IDs: 17
Enable/Disable SQL access without Attach/Detach: 25
Allow DBA to ATTACH database without password: 18
Exclusive DBE access not required for all functions: 46
Utility option to NOT Attach IMAGE Auto Masters: 64
Allow wild-card ADD USER in IMAGESQL: Utility: 24
ATTACH register referential integrity constraints: 22
Note: All eight of the above have been judged by the HP
R&D Database Lab to be either a "Medium" or a "Small"
effort to implement.
Another IMAGESQL Utility enhancement request was sent
in after the 1998 Ballot was voted on (credit to Ken Vickers):
If a key is split into separate components, it can no longer
be used for ODBC keyed access to TurboIMAGE.
IMAGE/SQL should allow the ability to reference and use
the original IMAGE key even if the key has been split.
COMMENT: Ken Vicker's suggestion sounds like a good
idea to me; and nobody has come up with an argument
against it so far that I have heard. And while the effort to
implement this idea has NOT been reviewed by HP, my
initial guess is that it is probably not a major effort.
Tony Furnivall initiated discussion and others have since
commented on another issue: IMAGESQL insists on a
maintenance word, even if none was specified by the data
base creator. Most people I've heard from feel this should
be classified as a bug and not an enhancement request.
Another view might be that if the database creator did not
specify a maintenance word, then s/he did not *want* anyone
to be able to access the database..... COMMENTS ??
=========================================
NOW THEN: Given that:
(1) Bundled ODBC direct to IMAGE was # 1 again in 1998.
(2) HP has said and continues to say "not going to do it".
(3) There are a *lot* of things that could be done to improve
the IMAGESQL Utility program; most if not all of them
are "Small" or "Medium" efforts to implement.
A COUPLE THOUGHTS / QUESTIONS / THINGS to PONDER:
(1) Even though overwhelming winner in the "Reasons" count
of why users voted for this enhancement was "Functionality", I
would expect that if most / all of the open enhancements
against the IMAGESQL Utility were done there would be
less demand for Item # 98I01. How much less I don't know,
but I do know that for large sites with complex operations and
lots of users, IMAGESQL administration can be a real pain...
And the more you need to run a 7x24 ODBC operation, the
bigger the IMAGESQL pain is likely to be.
(2) If HP is not going to bundle ODBC Direct to TurboIMAGE
anytime soon (I believe them), for those sites who can't or won't
purchase one of the third-party "direct" solutions improving the
IMAGESQL Utility would be a lot better than nothing.
THEREFORE, after all of the above right now I've only got one
specific QUESTION on this:
If most / all of the 10 IMAGESQL weaknesses / deficiencies /
omissions listed above were fixed, I would still want the number
one ranked Item 98I01 Bundled ODBC Direct to TurboIMAGE:
(a) Just as much.
(b) Noticeably less, but would still like HP to bundle it.
(c) A lot less (nice to have, but could get along without it).
One more note: Recent extensive testing by SIGIMAGE
Executive Committee (SIEC) member and SIGALLBASE
Chair Gary Biggs and others indicates that performance
of ODBC access to TurboIMAGE via IMAGESQL has
gotten a *lot* better compared to when it first came out....
Any other comments or suggestions above and beyond my
question are welcome. Please send replies if any to 3000-L,
so others can follow all discussion. Suggest retaining at
least the "SIGIMAGE '98 Ballot" part of the title if you want
to be sure that I notice it in collecting comments to forward
to HP.
BTW, for anyone who is interested in a very detailed and well-
thought out analysis and review of the IMAGESQL Utility and
its current deficiencies, I highly recommend the "IMAGE/SQL"
paper by SIEC member Fred White. It is available on the
ADAGER web site in six parts (either PDF or HTML format) at:
http://www.adager.com/TechnicalPapers.html
SIEC member Denys Beauchemin has also commented and
written about IMAGESQL issues, but at the moment I do not
have a ready reference to his work. If Denys has anything he
would like to add, I'll let him speak up for himself... :-)
Closing BTW: If you ask me any questions about this or any
other issues in the next eight days, be advised that I am only
going to be in the office Tuesday and Wednesday; otherwise
out of town or up in the mountains....
Ken Sletten
SIGIMAGE Chair
|
