I would like to know how one lazily loads repository items specified in ATG. I have tried all the options as given in the ATG Repository Guide under Lazy Loading Settings , but so far all the properties of an item configured to be lazily loaded are fetched along with the parent item.
For example if a User entity has one-to-many relationship with Organization, loading a user repository item results in all the organization items for the user getting loaded along with its properties such as billing and shipping addresses. This happens even if the organization collection is not used.
Any replies with some examples other than what is given in the repository guide or any gotchas would be great.
When a property has lazy loading enabled, the property is not fetch from the database along with the repository item. The property would be loaded from the database (with another SQL query) when it is needed ie. the property displayed in a page or used in a component.
Couple of examples when it make sense to use lazy loading properties:
When a repository has properties (quite often with large content) which are not used that often it is good to enable lazy loading on them e.g. you may have a "News" repository item type with properties as follows: title, creation date, content (long string). If you present list of news items (only title) ordered by creation date, it make sense to load only title and creation date. Therefore content property should have lazy loading enabled. That property will be fetched when you visit a particular news item. You do not have to load all the content properties during news list presentation which will improve page display performance.
Another example (similar to the above) might be with product catalogue and product description property. Usually in a category page you present only product title along with image. Product description is loaded and presented only when a product page is displayed. It make sense to set up product description property as lazy loaded in order to improve performance of category (or subcategory) page.
It is also good practice to use lazy loading if a query returns a very large number of repository items and you use only a few of them most of the time. The ones which are not used (they do not have to contain large content) can be lazy loaded.
I have been using lazy loading properties in all my ATG projects and I think it is a great feature. However, you have to consider all pros and cons before applying it. Sometimes it may cause more harm than good.
I hope that will help you better understand lazy loading and enjoy working with that feature in your projects.
Order History in ATG is a great example of when lazy loading can be useful. If you think about a typical use case for this it is along the lines of:
User views the order history page i.e. the list of all their past orders
Navigates through the results until they find the order(s) they are interested in viewing
Views an order or a small subset of the orders
Without lazy loading enabled ATG would load all the order objects for that particular user when viewing the summary of past orders. The order object a very large and complicated object, and loading multiple orders consumes significant resources on the database and the application servers.
With lazy loading enabled when the user views the order history page, only the information which is needed to display the list of past orders is loaded, anything which is not needed directly by this page is not loaded. If the user then clicks through to an individual order then any additional information is loaded at that point.
Related
For the first i want to sorry for my english. I'm learning domain driven design and trying to implement some concepts in an application i'm working on. My task is not so complex to fully implement DDD on all the levels but i really like it's principles and the core idea and try to use it.
Lets say app is selling books. So i have a Book entity and BooksCollection or BooksRepository. I'm working on frontend and that collection or updates to it is coming from server. And i want to represent it on BooksScreen in BooksList which consists of BookCard. Press on that card for the first calls something like selectBook which changes the selectedBookId in collection and for the second navigates user to BookDetails screen where the data of selected book is represented and the user can do some actions related to domain logic.
The first question is where do i put the loading state of that BooksCollection and according actions to change it ? Loading state is not a domain logic as i understand, it's not an entity status like "todo done" or something. But i need to show a loading indicator in the UI list when the collection updates, error for loading error and success respectively.
And the second is where do i put the the same loading state for single Book ?
I separate it cause for collection i may store that state in some application related class e.g. "BooksScreenState" or something with less stupid name. But what if i decide to show state for each specific card in the UI e.g. that specific card failed to load. Or i have a single User in app and his data can be loading, he can be authorized or not et cetera.
So i can summarize that to something like "how to connect domain and application logic to present a UI".
An interesting question. I don't usually think to apply DDD to the UI level because for the most part, the UI isn't full of business rules and also because I mostly use reactjs and UI frameworks are usually very prescriptive and don't allow much in the way of flexibility.
To answer your question though: If you did want to get a "loading" status in your UI (that's designed using DDD), it'd have to be attached to a "View" object or some representation of the "view"; because that fits the UL (ubiquitous language) better. Think of how you have described it in a little snippet of your question:
But what if i decide to show state for each specific card in the UI
That means your card is an object (entity or value object) that has a enumerable state of loading, loaded, error. You can queue on that object in your UI to display a desired representation of that field.
So, both your questions have essentially the same answer. Since you are loading the UI, your state is only relevant to UI objects and not entities that are in the domain model like "Books" that are represented in the backend. Even if you had a front-end representation of "Books" - like in javascript for example - having a loading state still makes more sense in the view object in the view layer.
Note that there's some simplification/flexibility to this answer because it's also valid for your design to have a View that's an aggregate that contains a Books. Those Book objects could have a "loading" state on them. All of this is still restricted to the UI layer though and such an aggregate and it's specific design will depend on the flexibility your UI-framework allows.
When our keen.io project was started, there were a bunch (hundreds?) of event collections accidentally created due to the names being changed dynamically (i.e. viewed page blog name X, viewed page blog Y, viewed page blog Z, etc). Does anyone know of an efficient way to delete all these collections (i.e. that does not involve deleting them one by one in the UI or via API)?
If I was able to query all the event collection names we have in our project, then I could easily loop through all the event collections and delete via the API, but I haven't found a way to get the event collection list back in a query.
Issuing a GET request to the Events resource returns schema information for all the event collections in a project, including properties and their types.
https://keen.io/docs/api/reference/#event-resource
keen-gem has an event_collections method that wraps this functionality if you're into Ruby.
It's best to loop through the schema one collection at a time as deleting many collections in parallel can lead to rate limiting.
(Using Ektron version 8.6.1)
Say I have a smartform ContactInfo, something like:
<ContactInfo>
<Name></Name>
<Email></Email>
</ContactInfo>
I would like to create another smartform (e.g. NewsArticle) and "embed" ContactInfo inside
<NewsArticle>
<Title></Title>
<Summary></Summary>
...
<ContactInfo>
<Name></Name>
<Email></Email>
</ContactInfo>
</NewsArticle>
My solution thus far has been to include a Resource Selector field to add a reference to an existing smartform instance. I would prefer to make the association at the configuration level, to make the data entry workflow more intuitive.
I'm using Bill Cava's ContentTypes and generating classes from smartform XSDs, so it would also make the presentation code more natural and type-safe in that embedded fields could be accessed directly (rather than having to make another request based on a reference ID, which may or may not be an ID to the smartform I'm expecting).
I gather this is not possible out of the box; I'm not opposed to hacking Workarea code to make something like this work. Does anyone have experience with a scenario like this?
I heard from an Ektron rep that they are planning on elevating the role of smartforms in an upcoming summer release - can anyone offer some more info to that point? Perhaps smartform composition like I've described will be supported?
Currently it isn't possible to do smartform composition. Depending on why/if you actually need a second smartform definition, you could just define the contact info in the news article.
If the contact info smartforms are related to the news articles in a one to many or many to many fashion, then using the resource selector as you have is the only way that I know of to create the relationship you are looking for.
If the relationship is one-to-one or many-to-one, then I'd suggest doing away with the separate smartform definition.
If you can clarify the workflow you are trying to achieve for the content authors, I might be able to respond better.
The Content Types would represent the data in the CMS. Suppose, as in your example, a NewsArticle contains a reference to a ContactInfo. Embedding the ContactInfo inside your NewsArticle might make sense from a presentation perspective, but it turns your ContentTypes into a one-way data model. You would lose the ability to construct a new NewsArticle and persist it into the CMS.
What might work well for you is to leave the content types as-is, with the id of the ContactInfo from the resource selector. Then create a NewsArticleDisplayModel... essentially a view model that contains the news article data plus ContactName and ContactEmail.
Now, if you need the contact info to be searchable, you could get really fancy with CMS Extensions and hook into the OnBeforePublish event to update searchable metadata with the name from the ContactInfo, so that the NewsArticle can be searched for using the values from the other "embedded" resource. That could get kinda tricky, though... ideally you'd have to also hook into the publish events of the ContactInfo objects in case something changes on that side, too. Then do you create a custom database table to track which NewsArticle content ids are using a particular ContactInfo?
Your solution can get as complex as it needs to, but I would keep the content blocks separate. If nothing else, you'll end up with a more maintainable and upgradable solution.
I have 2 1/2 years experience of VB.Net, mostly self taught, so please bear with me if I seem rather noobish still and do not know some of the basics. I would recommend you grab a cup of tea before starting on this, as it appears to have got quite long...
I currently have a rather large application (VB.Net website) of over 15000 lines of code at the last count. It does not do retail or anything particularly complex like that - it is literally just a wholesale viewing website with admin frontend, catalogue / catalogue management system and pageview system.
I don't really know much about how .Net applications work in the background - whether they are all loaded on the same thread or if each has its own thread... I just know how to code them, or at least like to think I do... :-)
Basically my application is set up as follows:
There are two different areas - the customer area and the administration frontend.
The main part of the customer frontend is the Catalogue. The MasterPage will load a list of products but that's all, and this is common to all the customer frontend pages.
I tend to work on only one or several parts of the application at a time before uploading the changes. So, for example, I may alter the hierarchy of the Catalogue and change the Catalogue page to match the hierarchy change whilst leaving everything else alone.
The pageview database is getting really quite large and so it is getting rather slow when the application is first requested due to the way it works.
The application timeout is set to 5 minutes - don't know how to change it, I have even tried asking this question on here and seem to remember the solution was quite complex and I was recommended not to change it, but if a customer requests the application 5 minutes after the last page view then it will reload the application from scratch. This means there is a very slow page load whenever it exceeds 5 minutes of inactivity.
I am not sure if this needs consideration to determine how best to split the application up, if at all, but each part of the catalogue system is set up as follows:
A Manager class at the top level, which is used by the admin frontend to add, edit and remove items of the specified type and the customer frontend to retrieve a list of items of the specified type. For example the "RangeManager" will contain a list of product "Ranges" and will be used to interact with these from the customer frontend.
An Item class, for example Range, which contains a list of Attributes. For example Name, Description, Visible, Created, CreatedBy and so on. The form for adding / editing loops through these to display relevant controls for the administrator. For example a Checkbox for BooleanAttribute.
An Attribute class, which can be of type StringAttribute, BooleanAttribute, IntegerAttribute and so on. There are also custom Attributes (not just datatypes) such as RangeAttribute, UserAttribute and so on. These are given a data field which is used to get a piece of data specific to the item it is contained in when it is first requested. Basically the Item is given a DataRow which is stored and accessed by Attributes only when they are first requested.
When one item is requested from a specific manager is requested, the manager will loop through all the items in the database and create a new instance of the item class. For example when a Range is requested from the RangeManager, the RangeManager will loop through all of the DataRows in the Ranges table and create a new instance of Range for each one. As stated above it simply creates a new instance with the DataRow, rather than loading all the data into it there and then. The Attributes themselves fetch the relevant data from the DataRow as and when they're first requested.
It just seems a tad stupid, in my mind, to recompile and upload the entire application every time I fix a minor bug or a spelling mistake for a word which is in the code behind (for example if I set the text of a Label dynamically). A fix / change to the Catalogue page, the way it is now, may mean a customer trying to view the Contact page, which is in no way related to the Catalogue page apart from by having the same MasterPage, cannot do so because the DLL is being uploaded.
Basically my question is, given my current situation, how would people suggest I change the architecture of the application by way of splitting it into multiple applications? I mean would it be just customer / admin, or customer / admin and pageviews, or some other way? Or not at all? Are there any other alternatives which I have not mentioned here? Could web services come in handy here? Like split the catalogue itself into a different application and just have the masterpage for all the other pages use a web service to get the names of the products to list on the left hand side? Am I just way WAY over-complicating things? Judging by the length of this question I probably am, and it wouldn't be the first time... I have tried to keep it short, but I always fail... :-)
Many thanks in advance, and sorry if I have just totally confused you!
Regards,
Richard
15000 LOC is not really all that big.
It sounds like you are not pre-compiling your site for publishing. You may want to read this: http://msdn.microsoft.com/en-us/library/1y1404zt(v=vs.80).aspx
Recompiling and uploading the application is the best way to do it. If all you are changing is your markup, that can be uploaded individually (e.g. changing some html layout in an aspx page).
I don't know what you mean here by application timeout, but if your app domain recycles every 5 minutes, then that doesn't seem right at all. You should look into this.
Also, if you find yourself working on various different parts of the site (i.e. many different changes), but need to deploy only some items in isolation, then you should look into how you are using your source control tools (you are using one, aren't you?). Look into something like GIT and branching/merging.
Start by reading:
Application Architecture Guide
So I'm using Core Data in an existing iPhone app, and I've set up two entities: Person and Item. The root view of my app's navigation (currently) shows a list of people, and you drill down from there to items. In short, the hierarchy looks like this:
Person -> Item
I want to add a new entity above Person in the hierarchy, called List:
List -> Person -> Item
Additionally, I want the user's first List to be created for them on startup, and for any People the user's already added to be assigned to that list.
I'm familiar with Core Data's lightweight migration & versioning feature, so I think I know how to add the new entity and relationship, but I'm not sure how to:
Create a List record on app start if they've never had the Lists feature before
Set all existing People records to belong to that new list.
One quick and dirty way would be to add some code to my app delegate's applicationDidFinishLaunching method that performs the migration by (1) checking to see if there are any Lists, (2) if not, creating the default one, (3) fetching all existing People from my data store, (4) setting each Person's list attribute to the newly created default list, and finally (5) saving those changes.
My question is: is there any faster or easier way to do all of that?
That's pretty much what you'd want to do. Use an NSFetchRequest to see if any Listss exist. If not, create one. Then do another request to get all the Persons. Here, instead of assigning the list property of each Person, I'd create an NSSet containing all your Persons and assign that to the List's people property. You did create an inverse property, right?
This is actually a pretty lightweight operation, all tolled, so I wouldn't worry too much about performance. Unless you've got hundreds or thousands of Person objects, your user will probably won't even notice.