Re: HELP I am about to quit on AS2005

Thanks for responding Craig,

I am working from a well prepared and relativly simple relational input
source as well, and there is no immediate time pressure for me either. That
doesn't mean however I feel a total lack of control and see myself faced
with software that does things that are highly unpredictable.

My issues with AS 2005 atm are:

* I miss good 'unfragmented' documentation.
* I miss Undo/Redo functionality in VS2005.
* I miss 'reverse engeneer' functinality for a deployed solution, so it can
me modified and/or learned from. This is a basic feature in all design
software, so why not here?
* Binding of dimensions and attributes that are designed from scratch, i.e.
those not yet present in a datasource. The source property is 'readonly' and
I just get stuck from there).
* Restore of backups does not always work, sometimes unexpected unspecified
errors pop up.
* The graphical representation of tables (with interconnecting lines) in
VS2005 is a joke. Try to arrange tables in a logical manner with straight
lines. There is an auto layout function I found, but I find it all to work
very unnatural and it usualy ends up doing unintended things such as forcing
loops instead of a staight line where even a child can see what my intenton
is.
* Due to various of the above, the wizzards generate either stuff I can't
turn off later or that I can't complete in order to make something work.
Hence the are pretty unsatisfying, it smells of feature creep, trying to
make comlex involving stuff a breeze, but at the expense of control on the
most basic level not to mention overhead. Not good!
* Making small changes somtimes cascade and change more then intended (see
dimension usage). Wihout undo/redo that means you have to do stuff over and
over again, and mistakes are bound to crop up more as things increase in
complexity (more dimensions).
* Attributes that do not form a hierarchy (even if one level), cannot be put
on any of the axes in an MDX query. They can be presented using properties,
but not in a way that allows reporting. For base tables this is pretty
wasteful and an unnececary restriction.
* Related to the previous point, I feel the MDX syntax needs more
streamlining. Right now its forgiving at times where I rather see an error
and very strict in an unusable way in others.
* Minimal changes in a model often mean complete rewriting of existing MDX
queries. The way things work realy promote sloppy and wasteful models whith
no doubt will eat performance when things need to scale.

In general I feel working on an unfinished product, based on a great idea,
but poorly implemented. I know there is loads more possible that can make
systems such as AS2005 shine, But I don't see why it couldn't have been
implemented better and it could certainly be done less vague and more
explicit about what the effects are when you change something. I don't feel
it supports 'in the field', 'out there' real live situations and practices
very well and that realy bites me atm.

I hope this feedback will provoke some answers, I seriously can't believe I
am the only one bumping into the issues listed.

Kind regards,


Peter


"craig_amtdatatechnologies@xxxxxxxxxxxxxx"
<craigamtdatatechnologiesdiscussionsmi@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in
message news:FAFEF419-CD74-47B8-9C21-A60531C53EC1@xxxxxxxxxxxxxxxx
Are there specific examples of things that are functionally defficient,
and
buggy?

I'm pretty new to AS 2005 myself. My approach to developing the cube
(storing 50 million lines data per year) has been to to gradually increase
the scope of cube (in number of dimensions), and gradually ramp up the
volumes (from 10,000).

The client is aware that I'm embarking on a learning exercise, so maybe
I'm
lucky in that I have time to find how to make AS 2005 work for me.

Having the staging database setup correctly, has certainly smoothed things
along for me (given the volumes of data concerned).

I'd be interested to here more of the specifics issues you've faced, as
others getting to grips with AS 2005 may well learn from these.



Regards,


Craig

"Peter de Heer" wrote:

My background is programming and hard-core SQL, and I am familiar with
several languages and handling of many types of data structures, both
conceptualy and in actual experience of over 20 years. I I expected just
by
the nature of the subject a rough ride from the start, but what I
experienced in the end was utter disapointment.

From my perspective the most basic things are extremely hard to configure
and even then needlesly limiting in use due to the way they seem
implementation. On top of that maintenaince and manual creation of
dimentions with sufficient control is lacking and and bound to end in a
dead
end road. The only thing that seems to work is to make a relational
equivlent of what you want to do and let wizzards come up with an
aproximation of what is intended.Which is nice for people not knowing
what
they want and it problably sells well too, but it does not solve business
problems or help developers create functional and maintainable solutions.

Now it could be the tools I use, so I will elaborate on that a bit. I use
SQL server manager managegement studio and MDX for data access. And I use
Visual studio 2005 to define the AS2005 side of my model. My griefes are
mostly with the later as it is full of features that do absolutely not
work,
has loads of eyecandy and unusable stuff while at the same time it is
functional deficient and buggy. This wouldn't be so bad if I could define
a
solution by means of the SQL server management studio, but that allows
only
viewing of data, not modelling. The next alternative is to use XML
definitions to create stuff, which for me is at shis point a step to big,
to
fast. And it requires far better documentation then is supplied IMHO.

The concept of cubes and how to define relationships, etc. is simple and
just a handful of basic rules (Q/A style) should enable anyone with my
background to get ahead and create solutions pretty fast in line with
design
intentions. I can't believe how little of that simplicity can be found
back
in the user interface and the documentation. Who are these tools designed
for? Given that its a developer tool I would say people like me, but I
sure
don't feel that way. This leads me to conclude that something te very
wrong
here, is the product even finished? Am I beta testing or participating in
some science experiment?

This is so frustrating...I wouln't even know where to begin listing the
issues I encountered so far without risking an all out flaming war! Not
to
mention it would take me a lot of time just to visualise the issues in a
way
that everyone can understand.

So please somone tell me what tools you use that give you the control
that
is needed, as I have written off visual studio 2005 for it. If nothing
else
pops up, I don't think I give it another look any time soon, which I
still
feel is a loss. But what I can't control even in the slightest just ain't
good enaugh to be used in a solution.





.

Relevant Pages

  • Re: Moving MsgBox
    ... you need to provide the same functionality which includes ... The taskswapper is a fully functional, ... necessary to use it as an OCX, and compiled the AX control. ... Memory footprint or combined file sizes? ...
    (microsoft.public.vb.general.discussion)
  • RE: [Full-Disclosure] Psexec on *NIX
    ... >> So you want all the functionality of psexec without installing psexec? ... >> You want to remote control any windows machine from your linux laptop? ... > not full control but just enough to start one program which would be run and copied on the targer machine only when the source machine has a username / passwd combo ...
    (Full-Disclosure)
  • Re: Drawing lines - another noob question
    ... It's a difficult question to answer regarding OOP. ... At the simplest level OOP combines the two so data and functionality exist ... Interfaces allow you to control polymorphism so that you don't create huge ... I was finally able to get the picturebox writing method to work. ...
    (microsoft.public.dotnet.languages.csharp)
  • Re: Preventing pasting in a TEdit
    ... What I said was the Delphi is ... >functionality do you claim to be able to provide in 1024 bytes of code? ... linked into each and every program for any control to work. ... >require the developer to know anything about Windows subclassing. ...
    (comp.lang.pascal.delphi.misc)
  • Re: Basic
    ... would be considered "extra" functionality in VB that is built directly ... For example, there are control arrays, ... REALbasic newsgroup about the new IDE. ... I had no pressing need at the time to target the ...
    (comp.lang.basic.visual.misc)