By using this site, you agree to our updated Privacy Policy and our Terms of Use. Manage your Cookies Settings.
435,595 Members | 3,721 Online
Bytes IT Community
+ Ask a Question
Need help? Post your question and get tips & solutions from a community of 435,595 IT Pros & Developers. It's quick & easy.

OO Design Best Practice - Please Help

P: n/a
Hi all, I have posted this in both the c# and asp.net groups as it applies
to both (apologies if it breaks some group rules).

I am making a web app in asp.net using c#. This is the first fully OO
application I will be making, also my first .NET application, so im looking
for any help and guidance.

Ok, my problems are todo with object and database abstraction, what should i
do.

I have defined certain objects my system will need, they are all linked.
Lets say these are

Patient --- Sample --- Sas

An Sas contains a Sample and a Sample contains a Patient. A patient can
exist on its own however a Sample needs a valid patient and an Sas needs a
valid sample.

At the database level i simply store 3 tables, one for each of the classes.
However, what is the procedure for loading my data back into objects before
I use them? I know I can make a static method within the class that loads
data, creates an object and returns it (although I was told this wasnt good)
or to use Data Access Objects (DAO)s as helper classes for my main classes
that do this work, again a good idea but i cant see it providing a solution
to one of my problems, see 2 below.

A few scenarios that are valid within the system:

1) I simply want to search the patients within the system. I could search by
patient parameters such as name and patientid. Will the patient class need
to contain code that performs this SQL query and returns an array of patient
objects, alternatively do I use a separate DAO (Database Access Object ) to
perform the patient queries? Now I need 2 classes to handle my patients (and
potentially 2 for Samples and 2 for Sas's etc etc).

2) I may want to search for all the samples created within a given date.
Remember each sample contains a patient. The actual SQL for this is simple,
I just need to perform a join, but where is the member function stored that
performs this SQL? How do i split the SQL to create Sample objects that
contain the correct patient objects?

3) The classes/objects will be viewed in a web form (i.e. i want to edit a
patient details, so I load the patient contents from the database into an
object which populates a web form) or in a pseudo data grid (i.e. searching
for a patient with the surname Smith and displaying all the results). I
didn't want to use a dataset as I would loose control over the appearance of
the search results/table displayed to the user. If I do use a dataset, there
is no need for an array of objects :S What am I to do?

These are the main issues im having developing the system. Coming from a
procedural PHP background, these issues seem very simple and easy to work
with, however being OO based makes them infinitely complex. The whole
running one query and the separate objects being created and nested
correctly it the real sticking point.

All advice and help appreciated. Please remember im a .net novice so if I
have missed a crucial facet of the technology, please do advice, i would
really appreciate it!

Thanks again, apologies for the long post.

Kind Regards
Taz
Feb 7 '06 #1
Share this Question
Share on Google+
4 Replies


P: n/a
Hello Taz,

I was once exactly where you are. I was coming from a PHP/PERL
background into the new, and quite intimidating OOP/OOD world.Here is
the best advice I can give you:

As your first OOP application, there will be a number of things you do
here that you will realize later were not done correctly.
Unfortunately, when I first start making OOP apps, it took me quite a
few tries before I finally got something halfway close to OOP.

I think the problem you are having here is you are focused in too much
on the actual implementation of your code. My advice is step back and
think of everything from a users standpoint. What I mean by this is,
what processes will the user be doing, what information will they want
to see, how will they access it, what will they be doing with it ocne
they get the data. It sounds like your halfway there because you know
what parameters you are going to need to do specific queries.

Are you familiar with UML at all? Many people here might not mention
it, but buy some UML books. This will help you IMMENSELY in the future.
I cannot stress this enough. Microsoft doesnt seem to support to too
much, but if you go to the java side of the fence you will see EVERYONE
uses UML prior to designing an app. I would also suggest some patterns
and practices books. After you get the hang of UML start to investigate
programming patterns, heres some good books: "Design Patterns
Explained" from Allan Shalloway and James Trott. Another very good book
in design patterns is "Patterns hatching" from John Vlissides

I'm getting a little ahead of myself here, i guess to answer your
original question:
I usually setup a database abstraction layer, create stored procedures
in SQL so all of my SQL statements are in the SQL server and not hard
coded in my app, then in your database access layer, setup overloaded
methods for all the different parameters you can accept. Next, setup a
business layer where you can process the data/enforce business rules on
the data (like checking for blank fields, missing fields, errors with
getting the data) and then a presentation layer for displaying your
information in ASP.NET.

Sometimes a business layer is a little overkill, but if your app is
large enough then I would think about creating one because it allows
for more scalability later down the road. I have a number of links on
OOP/OOD i can supply, just email me and i will be more than happy to
point you in the right direction. It is a long and tiresome road, but
once you get to a certain point, its like a lightbulb turns on and you
got it! (sometimes) :)

dkode

Feb 7 '06 #2

P: n/a
also,

search for posts on google groups with "dkode" as the author, you will
see some threads i have asked in the past, and a couple of them have
pretty good responses.

ta ta!

dkode

Feb 7 '06 #3

P: n/a
>
dkode


dkode, many thanks for the reply. I think my first post was a little
incorrect, im not new to OO, i have used it often and am competent, however,
this is the first major OO based web project i will be doing, including the
database stuff.

UML advice is great, I am lucky to have been taught this and developed class
diagrams of my problem scope. However, again this is where i am stuck, how
to define my class model based on how things are done in .NET.

As for the SQL trick, stored procedures do sound interesing, i will look
into them. However it still doesnt resolved problems of taking a result set
and then using certain columns to populate 1 object and other rows set to
populate a sub object.

Thanks again for your reply, i will definately be taking certain things on
board.

Kind regards,
Tarun
Feb 7 '06 #4

P: n/a

If you're using Sql Server, you can return multiple "ResultSets" from one
database call.

For Simplicity, I will use 3 Select Statements.
Select PatientID, PName from dbo.Patient
Select SampleID, PatientID, SampleName from dbo.Sample
Select SasID, SampleID, SasName from dbo.Sas

Ok.

You're looking for an OO way.
Create 3 classes. These will mimick your database Entities. (Actually,
your db mimicks your objects, but you understand what I'm saying)

Familarize yourself with the IDataReader object.
The way I do it is:
I create a stored procedure, which returns the 3 resultsets above. Then I
loop on them, with an IDataReader, to "Serialize" my data, into concrete
objects.
It goes something like this.

private object SerializeIDataReader( IDataReader idr )
{

while (idr.Read)
{
string patientName = idr.GetString(0); // where 0 is the ordinal number
of the select

Patient p = new Patient (patientName);
}

idr.NextResult();

while (idr.Read)
{
string sampName = idr.GetString(0);
Sample sam = new Sample(sampName);
}

idr.NextResult();

while (idr.Read)
{
string sasName = idr.GetString(0);
Sas sa = new Sas(sasName);
}

}

return someObjectWhichWrapsUpAllTheseThings;
}
That's a very abbreviated, and buggy code sample. But you get the point.

someObjectWhichWrapsUpAllTheseThings is something you write up.
The goal of SerializeIDataReader is that it can take any IDataReader.
Perhaps you create an IDataReader for 1 Patient, 100 Patients. Now, you'll
always have to have 3 ResultSets, but it gives you some flexibility.

..

Additional, you want to check out the EnterpriseLibrary
http://msdn.microsoft.com/library/de...tml/entlib.asp

This bundles up alot of database access stuff. Without reinventing the
wheel.
You saw I used IDataReader, the interface, and not any concrete
implmentation. This is so you can switch out your backend database, and not
really affect your code ... which takes an IDataReader and turns it into
concrete objects.

...

I don't know your setup exactly, but let me go with this generic sample.

Customer
Order
OrderDetails

In this situation, I'd have these 3 objects. I'd also have a collection for
each one.
CustomerCollection : CollectionBase
OrderCollection : CollectionBase
OrderDetailsCollection : CollectionBase

Customer
string CustomerName
OrderCollection AllOrdersForThisCustomer

Order
DateTime OrderDate
OrderDetailsCollection AllOrderDetailsForThisOrder
My stored procedure would have:

Select CustomerID, CustomerName from Customer
Select OrderID, CustomerID, OrderDate from Orders
Select OrderDetailID, OrderID from OrderDetails.

My serializer would do this:

1. Create a new CustomerCollection
2. Loop over the first ResultSet, creating Customer objects, and then
adding them to the CustomerCollection
3. Loop over the second ResultSet, creating Order objects, and adding them
to the specific Customer (matching up CustomerID)
4. Loop over the third ResultSet, creating OrderDetail objects, adding them
to the correct Order object.
5. return the CustomerCollection, which now has all data in it.

Because I wrote a generic serializer, I could use multiple IDataReaders.
What I mean , is that I could populate one IDataReader for All Customers,
and then another one for a Single Customer.

...

This takes some work. Sometimes a defer to the "poor mans" business object
of using a strongly typed dataset. But more recently, I opt for the OO
method.

...

see
http://spaces.msn.com/sholliday/ for a few tidbits. you may want to use the
freebie sql server 2000, called msde.


"Tarun Mistry" <no****@nospam.com> wrote in message
news:ea**************@TK2MSFTNGP10.phx.gbl...
Hi all, I have posted this in both the c# and asp.net groups as it applies
to both (apologies if it breaks some group rules).

I am making a web app in asp.net using c#. This is the first fully OO
application I will be making, also my first .NET application, so im looking for any help and guidance.

Ok, my problems are todo with object and database abstraction, what should i do.

I have defined certain objects my system will need, they are all linked.
Lets say these are

Patient --- Sample --- Sas

An Sas contains a Sample and a Sample contains a Patient. A patient can
exist on its own however a Sample needs a valid patient and an Sas needs a
valid sample.

At the database level i simply store 3 tables, one for each of the classes. However, what is the procedure for loading my data back into objects before I use them? I know I can make a static method within the class that loads
data, creates an object and returns it (although I was told this wasnt good) or to use Data Access Objects (DAO)s as helper classes for my main classes
that do this work, again a good idea but i cant see it providing a solution to one of my problems, see 2 below.

A few scenarios that are valid within the system:

1) I simply want to search the patients within the system. I could search by patient parameters such as name and patientid. Will the patient class need
to contain code that performs this SQL query and returns an array of patient objects, alternatively do I use a separate DAO (Database Access Object ) to perform the patient queries? Now I need 2 classes to handle my patients (and potentially 2 for Samples and 2 for Sas's etc etc).

2) I may want to search for all the samples created within a given date.
Remember each sample contains a patient. The actual SQL for this is simple, I just need to perform a join, but where is the member function stored that performs this SQL? How do i split the SQL to create Sample objects that
contain the correct patient objects?

3) The classes/objects will be viewed in a web form (i.e. i want to edit a
patient details, so I load the patient contents from the database into an
object which populates a web form) or in a pseudo data grid (i.e. searching for a patient with the surname Smith and displaying all the results). I
didn't want to use a dataset as I would loose control over the appearance of the search results/table displayed to the user. If I do use a dataset, there is no need for an array of objects :S What am I to do?

These are the main issues im having developing the system. Coming from a
procedural PHP background, these issues seem very simple and easy to work
with, however being OO based makes them infinitely complex. The whole
running one query and the separate objects being created and nested
correctly it the real sticking point.

All advice and help appreciated. Please remember im a .net novice so if I
have missed a crucial facet of the technology, please do advice, i would
really appreciate it!

Thanks again, apologies for the long post.

Kind Regards
Taz

Feb 8 '06 #5

This discussion thread is closed

Replies have been disabled for this discussion.