|
What you do is based more on what you posted here.
I can say for sure that attempts to "make my work easier" which result in dynamic/meta data solutions are always wrong. They do nothing but make the final solution MUCH harder to maintain.
That said there are numerous solutions that already exist for creating multiple layers based on actual data models. Those work when the actual concern is not a dynamic solution but rather the work involved in creating code for many data entities which is basically the same.
Depending on the solution it can create any or all of the following
1. The DDL
2. The DML - stored procs that act as an database API
3. DTOs and DAO in your language of choice
4. DAO API layers.
Myself I have been rolling such solutions myself for decades.
Only suggestion to the above I would make is that you must not allow the ease of use of the DTOs to allow you to extend their use into other layers of the application UNLESS they are free from all database hierarchy abstractions. And perhaps even then.
|
|
|
|
|
Ah this old chestnut again, every developer sooner or later wants to try and automate or pass this type of functionality to the user. As JSchell said it almost always ends in tears.
You do realise that nearly every one of us (the old farts here) has thought/attempted this type of project and yet most go back to single use code for each dimension table.
Moral of the story - don't ask the old farts for ideas, we are not happy with what we can think of and it will need a new approach to create something useful in this area.
Never underestimate the power of human stupidity
RAH
|
|
|
|
|
We have designed a 3-tier web app for a finance application.
The business tier is divided further into layers like manager, helper, util layers to modularize code and isolate the different functions i.e. core business vs non-business code from each other
Util layer has non-business functions which are required during a particular process e.g. DateUtils.java, EncryptionUtil.java etc
Helper layer has business logic which is specific to particular business process and not required in other business processes e.g. SomeThirdPartyInterestCalculationHelper.java, SpecificRequestBuilder.java
Manager layer has business process which controls the flow as well as implements some parts of business logic e.g. CustomerAccountManager.java has different methods for CRUD operations for customer account. It calls different helpers, utils, DTOs etc and gets the work done. It also implements some pieces of business logic. So,it performs mix of BPM role as well as parts of core business process logic.
As the processes become complex and lengthier, my manager layer is growing and does not look like well organized code.
I want seperate layers doing specific roles i.e. business process controller, business process execution (core business logic), CRUD operations which are DB specific, helpers (specific to processes), non-business logic
What can be a better design pattern to achieve this?
I am trying out Business Objects patterns to isolate different parts of business logic and coupling it with Application Service pattern.
So, for executing a business process, I would have:
1. ApplicationService - Would be a pure business process controller calling different business objects and controlling execution based on results of BO methods
2. BusinessObject1 - Core business logic in different methods - Called by ApplicationService
3. BusinessObject2 - Core business logic in different methods - Called by ApplicationService (if BusinessObject1 grows bigger or BusinessObject 1 and 2 can perform specific business functions)
4. IntegrationBusinessObject - To call other third party services required in business process
5. DomainEntityBusinessObject - CRUD operation for a particular domain entity required in process...will also have some business level checks required before or after CRUD operations
6. Adaptors - To convert formats for third party services - May be called by IntegrationBusinessObject
Idea is to make classes more compact and doing specific business functions. Also, control the process from a single class (Application Service) so that changing the process can be easier.
Is this design maintainable and scalable?
|
|
|
|
|
Quote: Is this design maintainable and scalable?
The "mental model" that you carry around of this particular app / system has no relation to the "user's" mental model of the business.
I can only assume you will never be actually talking to a "user"; since this is all plumbing.
The "application" programmers will love it (not).
Business processes are value "chains"; not "layers".
"(I) am amazed to see myself here rather than there ... now rather than then".
― Blaise Pascal
|
|
|
|
|
I have 3 DNN webservers behind a Citrix Load Balancer, the load balancer is configured for SSL Offloading.
I discovered that the login link doesn't work anymore. It just refreshes whenever it is clicked. The URL of the login link is: https://test.abc.net/User-Login?returnurl=%2f. The link when clicked supposed to take users to the page where they will login.
When I changed d Citrix load balancer to HTTP, everything works normal. I.e http://test.abc.net/User-Login?returnurl=%2f takes the users to the Login Page.
Any suggestion on how to resolve this issue will be appreciated.
|
|
|
|
|
Have you set the SSL Offloading header in DNN?
Setting the SSL Offload Header Value[^]
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
Thanks Richard. Your suggestion saved the day.
|
|
|
|
|
Thanks Richard. Your suggestion saved the day.
|
|
|
|
|
In an imagined situation we have a file without extension. We open the file with a hex editor. Is there any chance for us to interpret what we see in hex editor and ascertain whether the file has only instruction or only data (maybe as a text, maybe in some other format).
|
|
|
|
|
|
|
Hi,
I am trying to install PyQt on Mac but I cannot make it work. I used brew to install it
1 - "brew install pyqt".
2 - "brew cask install qt-creator"
Apparently it did everything it had to do but how do I start Qt?
Many thanks, Pedro
|
|
|
|
|
|
Hi,
I've installed with Brew and navigated to /usr/local/Cellar/pyqt/.../bin and there are 3 files there, but none of them start the Qt graphical environment. 
|
|
|
|
|
PyQt is a Python plugin to bind to the Qt framework. It can't be "started". It allows Python applications to use the Qt framework.
You have installed the Qt creator IDE (Integrated Development Environment). So you have to navigate to the installation directory of that and start it.
|
|
|
|
|
Ok will try that.
Brew installed pyqt, qt, python, sip and python@2. I guess then I should start it from qt directory.
I have already installed on my machine the "normal" Qt with C++. Can I use that installation and just add PyQt to it ?
|
|
|
|
|
Quote: I have already installed on my machine the "normal" Qt with C++. Can I use that installation and just add PyQt to it ?
What you have probably already installed are the Qt libraries, C++ header files, and also maybe the Qt Creator.
I would guess that PyQt uses your already existing Qt libraries.
An existing Creator should be also useable when configuring it for PyQt.
|
|
|
|
|
The problem is that brew puts everything in a directory called Cellar and installs everything there even if you have the program installed somewhere else.
For example, I have installed python on my Mac, nevertheless brewer installed it again under Cellar. Quite strange I think! So, given this, how can I "force" pyqt to use my original qt installation ?
|
|
|
|
|
Quote: So, given this, how can I "force" pyqt to use my original qt installation ?
I did not know the answer. But as already said I guess that it is already using the existing Qt libraries by looking in the common library directories.
It would also depend on the installation order when using brew. If you have installed Qt after PyQt, PyQt would not know where to look for Qt.
I have no Mac here and can't tyr it out. Why not just write a simple "Helle World" Python application using PyQt and check if it works. If so, you can try later to find out how to configure for using specific libraries.
|
|
|
|
|
Hi,
The simple Hello World type of program works perfectly on the Qt/C++.
Just an extra piece of information: brew, when installing software, checks for dependencies and installs them all.
On my case it has installed: python3, python2, SIP and qt. So I guess everything should work, as it is contained inside the Cellar directory.
|
|
|
|
|
I'm not sure wheter this is the right section to ask this. So please have indulgence.
Before some years when Software ran on a dedicated Hardware we could take a Fingerprint of the environement and make a "mas o menos" secure license based on the Hardware Fingerprint.
But now with virtualization, I'm lost.
Simply the question: Is there a way to take a forgery-proof fingerprint also from a virtual environment? I think no, but maybe I'm wrong.
Any ideas?
Thank you very much in advance.
[Edit]
It is more a question to secure to use specific protected data by a user, but I think in the end it remains the same Problem like Software licensing[/Edit]
It does not solve my Problem, but it answers my question
modified 19-Jan-21 21:04pm.
|
|
|
|
|
|
Thank you very much for this.
Quote: (You can detect if you are running "virtual").
I'm Aware of this, so I can decide whether to allow virtual or not.
Quote: There are "hardware fingerprints" available even in VM's (apparently).
That is not clear for me how this should work. I need to investigate more on this.[Edit]In case you mean a Fingerprint of the of the underlying machine I see a Problem:
a.) Some of our customers do move the virtual machines frequently from one physical Server to another (because of securtity reasons, whatelse? I don't know it exactly).
Quote: What about Azure?
No idea about it at the Moment and I think also not a solution for some of our conservative customer.
Thank you a gain very much.
It does not solve my Problem, but it answers my question
modified 19-Jan-21 21:04pm.
|
|
|
|
|
I think the way to go is "always connected".
Unless it's "field equipment", I expect every sane person to have internet.
Put your key services in the cloud, and "license" the client.
Your back end will know at all times which "license" is accessing from which IP; and how many "instances".
You can now license on an "instance" basis. Maybe even compute "overhead".
And there's the "token system". Issue tokens that expire and have to be renewed by "calling home". The same client renewing the same token is obviously a flag.
"(I) am amazed to see myself here rather than there ... now rather than then".
― Blaise Pascal
modified 20-Apr-18 19:02pm.
|
|
|
|
|
That was my idea, but Management....
Thank you very much again!
It does not solve my Problem, but it answers my question
modified 19-Jan-21 21:04pm.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
