|
I don't totally agree here. Implementation aside, if you were to draw me an ERD then you would agree that Widgets have an Owner of type Person. A person owns none or more widgets. Regardless of whether you have a ASP.Net/Webforms or command line, this relationship exists.
You business layer should also have Widget.Owner, and likely Person.Widgets[]. Sure, your database reflects this as a foreign key. Thats fine. Call it a coincidence. Your BL is based off the ERD, your database is also based off the ERD.
They do both represent the same things, expect them to be similar.
|
|
|
|
|
Mark Churchill wrote: You business layer should also have Widget.Owner, and likely Person.Widgets[]. Sure, your database reflects this as a foreign key. Thats fine. Call it a coincidence. Your BL is based off the ERD, your database is also based off the ERD.
Mark, but this leads to a circular dependency, i.e. a Person has Widgets collection, each widget has an Owner, which is a Person, which has a Widgets collection, each widget has an Owner, which is a Person, which has a Widgets collection, each widget has an Owner...
You know what I mean? 
Currently my idea is to have a User class, which contains WidgetList (inherits from List[Widget]), but Widget only has UserId as its owner, not the whole object. If you need the owner, you can retrieve it from UserManager (another class which manages objects of type User). This, of course, means that ASP.NET databinding is not so easy, because I need to add some TemplateFields and fill them in in GridView event handlers.
Kind regards,
Pawel Krakowiak
|
|
|
|
|
That circular dependancy seems to be correct behaviour to me. What is important is that the User and the Widget are the same instance every time around.
I would expect "somePerson == somePerson.Widgets[0].Owner".
This is similar to say a heirachy of Controls on a page. "someThing.Controls[0].Parent" could have you going in circles in a similar manner.
Diamond Binding will give you a Owner property on the Widget class of type Person, and an IList<widget> on the Person class. The actual implementation of IList depends on whether you have picked lazy-loading. You could also configure an IDictionary<x,widget> where x is a property of Widget that is unique per Person.
Think of it this way: Is it incorrect of me to ask "Who is your father's son's father's son's father?". It might be inefficient of me to ask that, but the query has an answer in your problem domain.
|
|
|
|
|
So, you suggest to have an accessor like this:
<code>private int _ownerId;
public User Owner
{
get
{
return UserManager.GetUserById(_ownerId);
}
set
{
_ownerId = ((User)value).UserId;
}
}</code>
User = Person, I currently have that class named User, although it conflicts with a class from .NET Framework sometimes, so renaming it to Person is a good idea.
This is a simplified example, but something like the above? This would allow to bind easily the property in ASP.NET, like "User.Owner.FullName", but still wouldn't endlessly load objects, right?
BTW. Is DiamondBinding an ORM? I have currently a license for EntitySpaces[^], but am not using it at the moment. The way how ORMs present the objects looks like more fitting for the DAL rather than BLL... But this is an another discussion. 
Kind regards,
Pawel Krakowiak
|
|
|
|
|
Yep that seems ok. If the UserManager has a caching strategy of some sort then you won't have any problems with circular references giving you duplicate objects either. (Although theres a redundant cast in that setter ).
You could also keep a local cache of the child object - this would be more efficient when pulling out say User.Owner.FirstName + User.Owner.LastName - it would avoid two trips to the UserManager. Of course with any caching you have stale-cache issues to solve
As for lists, you can have your Widget collection implement IList<widget>, but not actually retrieving the objects until iterated or an item requested (lazy loading again).
Diamond Binding is an ORM and definition generator. It follows the ActiveRecord model, so its less data-model oriented than other ORMs - so you are mainly dealing with objects at a business level: Customer c = Customer.Find(primaryKey); IList/Customer/ l = Customer.FindAll(); (Edit: Forum broke my generics!) Theres an example on the product page here[^]. Give the Personal Edition a try if you like, theres no functional difference between it and the Professional / Enterprise Editions. We aim to handle the entire data access requirements and just let you build business logic on top of the model we give you.
|
|
|
|
|
Mark Churchill wrote: If the UserManager has a caching strategy of some sort then you won't have any problems with circular references giving you duplicate objects either.
I am not sure how to do this yet, can you point me to some articles?
Mark Churchill wrote: Although theres a redundant cast in that setter
Yes. I thought value was of type object, forum doesn't give me any guidelines. 
Thanks for all your comments so far.
Kind regards,
Pawel Krakowiak
|
|
|
|
|
A really simplistic implementation would be to add the caching to your UserManager.Load(object pk). Check the cache, and if its in the cache then return it. Otherwise load the object's data and add it to the cache.
You can use a Dictionary for your cache for prototyping. ASP.Net provides a cache (I think you can safely use this outside of the web framework too) that you can add objects to with a key and an expiry.
You might want to add some way of forcing a reload of an object as well. To be honest though if you are writing your own data layer you are going to have to draw the line for features somewhere. I haven't done a KLOC count on Diamond Binding for a while - but I'm guessing its approaching the millions, and thats not including NHibernate
|
|
|
|
|
Slightly OT... But I was wondering, how does Diamon Binding differ/compare to the ADO.Net Entity Framework coming up in VS2008?
Me: Can you see the "up" arrow?
User:Errr...ummm....no.
Me: Can you see an arrow that points upwards?
User: Oh yes, I see it now!
-Excerpt from a support call taken by me, 08/31/2007
|
|
|
|
|
I guess the cheeky reply is that Diamond Binding is avaliable now, and is mature, and leveraging Hibernate, which is also decades mature. Last I heard the Entity Framework was dropped from Orcas and was going to be released "out-of-band" sometime in the first half of 2008.
EF, like a lot of ORM products tends to be very heavyweight and favours model-first development. It is a very complex product, however I think the key feature that everyone is interested in is LINQ to Entities. LINQ is not really EF specific. After the product becomes more mature, we will look at including LINQ support in Diamond Binding. Although we find that Expressions and HQL support cover most use cases sufficiently at the moment.
DB is designed with simplicity in mind, and is geared towards keeping your business objects synchronised with the database schema, and handling everything in between for you. If the schema changes, its a one-click synchronisation. One aim is to stop developers getting bogged down in mapping, and just let them add business value to the application!
|
|
|
|
|
Mark Churchill wrote: A really simplistic implementation would be to add the caching to your UserManager.Load(object pk). Check the cache, and if its in the cache then return it. Otherwise load the object's data and add it to the cache.
Sure, but I am concerned with the stateless behavior of web application / HTTP. As far as I know when the page is pushed back to the calling browser, it is disposed of and doesn't exist anymore, hence I am not sure how to cache data, otherwise than in some process which runs all the time on the server. Honestly, I haven't been playing with cache mechanisms of .NET yet, it's something on my list.
Kind regards,
Pawel Krakowiak
|
|
|
|
|
.Net is rather nice and makes it fairly easy for you. I found this article when I googled for the MSDN pages for Context.Cache, and it seemed easier to read anyway:
http://www.developer.com/net/net/article.php/1477771[^]
The guy does it a bit oddly though, giving a server too busy when it can't find the cached item (I'd go for a GetQuotes method myself that checks and updates the cache). HTH
|
|
|
|
|
I'd would like some suggestions on how to implement the actual bands, or groupings, editor GUI for a file export system I am busy with. Currently, I have a tree view on the left, explorer style, which lists as top level nodes all the Sections in an export, e.g. Customers, Clients, Customer Invoices etc. Then, under each section I have the data source queries for that section, and all record defintions, i.e. column collections, for that section.
Now in the XML definition file, under the record definitions, I have groupings to one level, where each grouping may be on Invoice Number. Each grouping can have a header, emitted once before each group, body emitted for each item in the group, and a footer, emitted once after each group. Header, body, and footer can contain one or more lines that correspond to record definitions. I'm now looking for ideas on how to present the layout of these groupings to users for editing.
|
|
|
|
|
Hi everybody,
I have 8+ years experience both in Java and .Net. Currently working a Software Engineer. I am getting opportunities for .NET architect position. But I am not aware of what kind of skill set required for this position so that I can upgrade my skills.
I will be thankful, for your suggestion and information.
Intelligence is measured by common sense not by how many scholarly books you read.
|
|
|
|
|
M LN Rao wrote: .NET architect position. But I am not aware of what kind of skill set required for this position
Probably the most important is Buzzwords. Seriously, if they don't provide a definition for architect it could mean just about anything.
|
|
|
|
|
Generally speaking an architect needs to be familiar with the most common ways to design information systems. On the "system" level, that means knowing the differences between desktop apps, client-server systems (including web applications), and distributed systems. On the application level it means defining the tiers and choosing the appropriate paradigm (e.g. service oriented or object oriented). It may also mean selecting technology - the term ".NET architect" is a bit odd imo, as one should normally choose a technology based on the desired architecture rather than the other way around. Nowadays an important consideration nearly everywhere is "interoperability", the ability to integrate with other systems, and I'd recommend that you study WCF if you don't already know it, as it is a very useful technology and easy to use with .NET.
|
|
|
|
|
I saw in many CRM products allowing there customers to add Fields at implementation level
Example: Customer of CRM Product can add “Phone2 “ as an extra field in to Registration form
Can any buddy tell me how this type of Architecture design will work?
|
|
|
|
|
|
Thank you Gary
But my question was many CRM Applications giving functionality to client to add controls/fields to product without knowledge of vendor like Outlook forms you can add your own fields
--
Regards
Pritesh
|
|
|
|
|
Yes. The articles I pointed to in my previous answer show you how to allow customers to store their specifically required information. You can then query the database and generate fields (at runtime) to supply the required information.
--
Cheers,
Gary
http://www.garyshort.org
|
|
|
|
|
but how we will query those fields without our knowledge mean to say we do not know which fields they created.
--
regards
Pritesh
|
|
|
|
|
The following T-Sql code will return the names of all the columns and their data types for a given table (in this example the table is titles from the pubs database). You can then use this to add controls at run time.
SELECT
c.name,
t.name AS [Data Type]
FROM
sysobjects o
INNER join syscolumns c ON c.id = o.id
INNER join systypes t ON t.usertype = c.usertype
WHERE
o.name = 'titles'
ORDER BY
c.name
--
Cheers,
Gary
http://www.garyshort.org
|
|
|
|
|
With an emphasis on providing practical advice, Code Generation 2008 is the ideal opportunity for architects, developers and others to understand how to benefit from emerging tools and technologies in the broad area of Code Generation.
Call for Speakers:
Submission Deadline: Friday January 18th 2008
We are currently seeking high-quality session proposals covering topics in model-driven software development (including Software Factories, Model-Driven Architecture (MDA), Domain-Specific Languages (DSLs), Generative Programming, Software Product Lines and related areas).
Sessions could cover topics such as:
- Tool and technology adoption
- Code Generation and Model Transformation tools and approaches
- Defining and implementing modelling languages
- Domain Analysis and Domain Engineering
- Language evolution and modularization
- Meta Modelling
- Runtime virtual machines versus direct code generation
Real-world case studies based on any aspect of these and related approaches are particularly encouraged although more theoretical sessions are also welcome.
Take part in Code Generation 2008 and find out why industry observers think that Code Generation is the next evolutionary step in Software Development.
Accepted speakers have their conference fees waived.
Code Generation 2008 takes place in Cambridge, UK from 25th - 27th June 2008
For more information on proposing a session please visit:
http://www.codegeneration.net/conference/speak.php
What people said about our previous conference - Code Generation 2007:
"I've been working in domain-specific modelling for a dozen years … and in this time this has been the highest-quality conference on this topic that I've been to - and I've been to a few."
"The combined—for that matter, individual—expertise present was remarkable, and presented a tremendous opportunity for knowledge exchange."
"The presentations were all top quality, making it often difficult to decide between the concurrently running sessions. The wealth of MDD knowledge present at the event was impressive, not only from the presenters, but from the other delegates as well."
Mark Dalgarno, CG2008 Organiser
Writing (mainly) on software at The Variation Point
|
|
|
|
|
Do we need to include activity diagrams, when we create Software Requirement Specification report?
Thank you,
Chatura Dilan
|
|
|
|
|
How long is does a piece of string need to be?
Same kind of question.
The answer is really 'it depends'.
|
|
|
|
|
I'm developing a simple order capture application in MS Access for a client that doesn't know what they want, beyond this very simple description. They must be able to maintain a list of products, capture and print quotations, convert quotations to orders.
This is far from my normal work, which seldom involves any GUI work, or very little. If I compose a feature set, it's what I would want, as a developer, and too much for the size of the job. I'm trying to find a set of guidelines or standard features that such an application should have, in order to present a tidy application.
Where can I find these? Should I maybe use a generator such as Iron Speed, whose designers know all about this already?
I do not believe they are right who say that the defects of famous men should be ignored. I think it is better that we should know them. Then, though we are conscious of having faults as glaring as theirs, we can believe that that is no hindrance to our achieving also something of their virtues. - W. Somerset Maugham
My New Blog
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
