Re: B-Tree Index Usage

Hi Tony
I think that you have missed the point.

The translate is documented and it does work in the VME. The major problem
is that it does not work in the FSI.
The second problem is that A correlatives as documented with mult-values do
not work with indices. This has the secondary effect of making one wonder if
anything else has been mucked up.

I am delighted that Ross Ferris's testing means that he recommends it. On
the basis of this I will do so to my clients who I insist maintain Gold
support levels.

My request for comment on the Re-Index a single item command has revealed
that people like yourself dont use indexing. In my opinion without the
ability to do translates it is of limited value and this simple command
could fix that. I would be fascinated to hear of how you would work-around
it without re-indexing the entire file every time.

As for being an unnecessary change I am again flabberghasted. Many people
have been using b-trees in Pick for over 20 years because they are an
essential tool. We would like them to work usefully at the system level,
however until they are fixed we have to avoid the D3 implementation and do
it ourselves. As Tom Del. says this reduces the value of the product in the
eyes of the outside world. By the way I was one of those expert double
de-clutchers that he speaks of driving trucks and tourist coaches in Europe
and the ice covered slopes to New Zealand snow fields. I now sit and watch
in amazement as they just flick the switch whereas we had to not only
double-declutch we also had to change between high and low range at the same
time. I do think that in some ways we D3'ers are still back in that era and
would love to just flick the switch rather than rolling my own.

Peter McMurray

"Tony Gravagno" <g6q3x9lu53001@xxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:scsod2dhunm1re8c8bsre8tm6f8hv9dgpk@xxxxxxxxxx
"Peter McMurray" wrote:

Hi Tony
It is a pity that you did not read the original post.
I asked for feedback on the need for a re-index feature that would make
the
indices far superior in my opinion. I would still appreciate your
opinion
before I contact Raining Data through Tdata.

I neither like nor dislike the idea. Off-hand I'm not enthusiastic
(not that it matters) because I don't consider indexing on a
translated value to be a currently valid function. At least to my
knowledge it's never been a documented/supported function, so I can't
support an enhancement request that allows this sort of thing to be
treated like it is supported. I don't know if that makes any sense
but I'm sure people will give me crap about it, so I'll restate it.

Feature: system-level updates of index to translated values
Example: A; (3:"*":4)(Tfile;x;;1)
Proposal: RD should write a re-index feature that would modify the
index with new end-result values if atb1 in the translated file
changes.

If this were my software, my response would be that I wouldn't want to
create an add-in for a feature that wasn't supported in the first
place. The proper solution is to simply support the feature. Until
that decision is made it doesn't make a lot of sense to build around
it.

Also, you don't need RD to write this. This sort of "back-indexing"
can be done with application code until RD decides to support the
desired functionality. This is where we get into the discussions
where I don't think it's appropriate to bug vendors to do something
that we can very well do ourselves - I'd prefer that they work on the
things which we can't do in the field. Similarly I think RD puts a
lower priority on changes which can easily be implemented in the
field, but if they hear about it often enough they occasionally build
stuff in - which is why there are several versions of different verbs
in the MD.

Now, is the feature described above valid? IMO, yes. But this is
where I maintain that D3 has worked without this feature for many
years, and if enough people actually wanted the feature, they should
have been freakin asking for it like I've been saying for so many
years. The wrong time to get indignant about a feature being missing
is after a release goes production that's been in development for x
years. What I see in this thread is one person saying he wants a
feature, and bunch of other guys saying how stupid it is that the
feature isn't already in there. That's politics talking, it's not
concensus from your fellow end-users. Obviously no one has presented
a business case to the vendor to support the feature. I don't know
about you but if this were my software that we were talking about, I'd
be a little ticked.

And frankly when I was at RD and trying to get you guys to petition
for features that you wanted, I _was_ ticked that I couldn't get
enough backing from the field for stuff that we've all actually wanted
for years. Mark and I had discussions where I would maintain that
there was demand for features even though RD didn't get official
requests, and he would maintain that if no one asked then it wasn't
important enough to make a change. Despite my eagerness to get some
changes implemented, from a business position I had to agree with that
logic. You guys only make it easier for a company to take such a
position.


As for the Date issue if you check you will find that 7.5 does NOT use
the
date index unless it is limited as I said in my original post.

I'm sorry, but I wasn't even thinking about this in these exchanges.
I've said before that I don't even really use D3 indexing so I don't
really care about the implementation details.


I am disappointed that you regard a release with major indexing faults as
acceptable. Funnily enough I don't,

There is a difference between faults, and lack of support for
desirable features. When indexes don't work from the update
processor, that's a fault. When a new release comes out that isn't
any different than it's last 15 predecessors, that means a conscious
decision has been made to not support the feature.

Please ask TData and other beta sites if they consider this to be a
major fault. Then ask if they reported it as such during the first
beta, the second beta, or maybe the original beta three years ago.
These are your vendors, tell them what's important, a public forum
isn't the right place for this.

As far as what's acceptable, I had my chance as QA Manager and DBMS
Product Manager to convey my definitions of the term "acceptable". I
found my definitions often didn't agree with those of Marketing,
Sales, or Engineering. Not surprisingly, my definitions did often
coincide with those of Support. In my mind, something unacceptable to
Support (or the field) translates directly to something that should be
unacceptable to all other departments. As evidenced by their great
loss of market share, PS/RD has never made this mental connection.
Suffice to say my opinions and the way I expressed them seemed to have
made me more of a liability than an asset. These days I see RD as a
vendor with limited resources and as a user/developer I want them to
focus on the things that I think are the most important, so I try not
to bother them with things that aren't higher up the priority list.
The word "acceptable" tends to get redefined depending on which
vendors we're talking about and what else we need from them at the
moment. Maybe another term for this is "pick your battles wisely".


and furthermore if a new release of
documentation is wrong on the subject of how to do an A correlative I
find
that a disaster for product presentation. Again I say these are NOT new
features.

...and this isn't my battle. The docs are far better than they used
to be. Most people today are using AP/Ref or the D3 v7.1 Ref which is
about 6 years old. As of now we have a completely new doc baseline to
pick apart. My recommendation is that you make a list of doc issues,
be very clear about what's wrong and what should be there, then send
your list periodically to your support provider. Get an action item
number that confirms receipt of your changes, and hope that your
changes are reflected in the next doc version. What else can I say?


I am particularly intrigued as to why you think everyone should be left
to
fall over these things on their own rather than discussing them.
Particularly as you were saying only a month or so ago that the Pick
classes
should not be used by anyone. Are you the only one allowed to criticise
:-)

Criticism is fine and constructive in many cases. I've just been
taking issue with your position that this indexing thing is a
high-priority product defect that needs to be addressed immediately.
File an enhancement request, petition our colleagues to do the same,
but until your change is implemented just live with the software as it
has existed for the last 10 years.

And I'm sorry, but I don't know what you mean by "Pick classes". Do
you mean D3 Class Library, MVOO stuff, certification, something else?

I will refer you to one of my clients attitude to software faults "If I
buy
a truck I expect to turn the key and drive off. Why should software be
any
different?"

We aren't talking about something as critical as an engine or
headlights which we expect to be present if we're going to operate a
vehicle.

If that truck never had a satellite radio, your client never asked for
a satellite radio, and no one else driving this popular model truck
seems to care about a satellite radio, how would you feel if your
client insisted that the manufacturer add a satellite radio to the
entire line of trucks before they drive off with one? My attitude?
Drive the truck, buy a radio, and we'll consider adding it if more
people seem to want the option. And if the client has already been
driving this truck for years, I'd ask them why this radio is now so
important that the truck can't be driven without it.

Healthy discussion if a bit "warm". Thanks Peter.

Tony


"Tony Gravagno" <g6q3x9lu53001@xxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:lg9ld2tr0utfhtl29gqtlircf0ovoud74p@xxxxxxxxxx
"Peter McMurray" wrote:

Hi Ross
All my remarks are directed at 7.5 although I have checked some of the
results back against 7.4.4.

Ross is right, v7.5 now supports date ranges using:
GE x AND LE y


About the rest of it, I'm trying to avoid this discussion because my
response is the same for pretty much every posting. If the software or
docs are wrong, email your support provider for a change or
clarification.

BTW, note that v7.5 is free to anyone on a support agreement. This
release was originally D3 8.0 but many of the fixes and enhancements
were pulled back into v7 to make them available without an extra
purchase. I look forward to working with a great many of these
features and I applaud RD for this particular business decision.

If you don't like what you have now, considering that at least you're
getting it for free. It could have cost you a lot more to discover
all the new bugs and other things you don't like. :)



.

Relevant Pages

  • Re: she might rip close, unless Rifaat upsets demonstrations in Maifys flow
    ... My convincing pool won't make before I feature it. ... When doesn't Vance exceed downstairs? ... Almost no respectable mls with regard to the crucial guerrilla were ... it indicates a discharge too red in support of her ...
    (sci.crypt)
  • Re: For Each
    ... the counter is a counter for the iteration - not the group ... I think the feature would be useful. ... Framework and having a counter in a for each loop, ... I meant the necessary changes to the framework to support: ...
    (microsoft.public.dotnet.languages.vb)
  • landing alongside chosen ulcer
    ... launch commonly if Mhammed's feature isn't tricky. ... in support of the gigantic harbour. ... searchs the sure wish under the television? ...
    (rec.music.christian)
  • Re: Is there no Good image viewing and managing software ?
    ... feature and what is functionality.. ... is great because it has universal support. ... you should be able to drag ... while I select files - I want to go into tag mode and then click away ...
    (rec.photo.digital)
  • Re: Date(Y,M,D) feature or problem?
    ... Bug or feature? ... It translates 31 April 2007 to a real date, which can be very useful in certain ... Prev by Date: ...
    (microsoft.public.excel.worksheet.functions)