Potassium Problems posted:

Are you talking about an open generic? like typeof(List<>)

NihilCredo posted:

List<T> is a parametrically polymorphic type.

The process of assigning a specific type to T, e.g. String, is sometimes called the monomorphization of List<T>.

So List<String> is a monomorphic type, but so is String and every other non-generic type, so it's not a distinctive term. I suppose the correct term would be "monomorphed", although I've never seen it used.

Chrungka posted:

How about using genericType for closed generic exclusively and genericTypeDefinition for open ones. It might not be 100% correct, but at least it's consistent with .Net reflection naming.

code:

var genericType = typeof(List<>).MakeGenericType(typeof(string));
var genericTypeDefinition = typeof(List<string>).GetGenericTypeDefinition();

quote:

It is important to note that a method is not generic just because it belongs to a generic type, or even because it has formal parameters whose types are the generic parameters of the enclosing type.

quote:

The common language runtime considers nested types to be generic, even if they do not have generic type parameters of their own.

Red Mike posted:

There is a page with terminology in the documentation. "Open" isn't explicitly used in there, but it's what the rest of the documentation tends to use. I think the same terminology is used for method definitions, however that isn't really explicitly stated.

I recommend you read this: https://docs.microsoft.com/en-us/dotnet/csharp/programming-guide/generics/generics-and-reflection
Followed by this (linked from the previous one), this one has the breakdown of terminology despite it being on an API ref article: https://docs.microsoft.com/en-us/dotnet/api/system.type.isgenerictype?view=netframework-4.8#remarks

Rocko Bonaparte posted:

I think I saw "open types" and "closed types" being used throughout so I'll see if it makes sense for me to use those terms instead. If there's one thing I don't like after using monomorph is that polymorphic as an alternative implies run-time abstraction.

It also looks like I should use ContainsGenericArguments instead of IsGeneric/IsGenericDefinition.

Boz0r posted:

I've moved over to generating client proxies with NSwag in MSBuild, but my XML comments aren't included in the swagger definition. How do I fix this? I'm using the Web API via reflection swagger generator.

NihilCredo posted:

Disclaimer: I'm using the NSwag middleware so I don't know if the generation via NSwagStudio + reflection even supports XML comments.

Are you outputting the .xml files along with the .dlls? It's a project option to publish those, and the XML comments are printed there.

fankey posted:

I'm trying to support user supplied custom fonts in my application. The fonts will be embedded in the document as a byte[] and it can't be required that they are installed by the regular Windows font management. I need to create a Typeface based on the font which will be used directly via DrawingContext calls as well as assigned to regular Controls. As far as the user is concerned the fonts will be specified using the same syntax as the CSS @font-face directive where the src() is the file name used when the font was embedded in the document. I'm running into 2 issues-

1. WPF really wants the fonts to be either resources or files on disk. I haven't figured out a way to make it load from a byte[]. The best I've found is saving to a temp file which I don't like doing if it's not absolutely necessary.

2. I can't figure out how to load the font via filename. Even if I pass in a file to Fonts.GetTypefaces it returns all the fonts in the same directory as the file with their names resolved to the names as defined in the font files themselves. This would be great for populating a dropdown with the available fonts but that's not what I'm trying to do - I want to get the Typeface based on the filename, not what's defined in the font file. For added confusion if you just pass in a directory into GetTypefaces nothing is returned but if you can pass in a file that doesn't exist as long as it ends with .ttf and it'll find every font in the folder.

EDIT : looks like I can create a GlyphTypeface based on a specific file but I'm not seeing a way to get a Typeface from a GlyphTypeface.

EDIT2 : I've come up with what seems like a possibly horrible and fragile solution

• Save all the byte[] to temporary files in a given path
  
• load the GlyphTypeface based on each file name
  
• from the glyph get the the name of the FontFamily and create a family->path dictionary
  
• use Fonts.GetTypefaces(temppath) to load all the typefaces
  
• look up the path based on FontFamily, now create a path->Typeface dictionary
  

In my extensive testing of 2 fonts it seems to work but I'm open to any other better ideas that don't seem as hacky.

raminasi posted:

This solution just goes around WPF to load the font family, which might look kind of scary but should work fine in practice. Have you investigated that kind of approach?

fankey posted:

Yeah I tried that approach and I couldn't get it to work - all the fonts ended up being Arial. It's possible I was doing something wrong - in my case I need to end up with a Typeface not a FontFamily which is what that approach accomplished. Even if it worked I'm not sure that approach would allow me to differentiate between different weights and styles of the same Family that came from different font files. I'll play around more with it since if it works it's a lot better than saving out files to disk - especially with awesome bugs in GlyphTypeface that requires every font be in its own directory.

raminasi posted:

With the enormous caveat that I've never done anything like this myself, is this TypeFace constructor not what you want? Or does that always yield Arial?

Mr Shiny Pants posted:

Create a Guid and stuff it in the headers and pick it up on the receiving end? That's what I used in the past.

EssOEss posted:

I can't help you but wow, Azure Media Services is still alive? I remember when they fired everybody but like 1 developer. What do you use it for? What are its strengths?

Boz0r posted:

You're right, of course, but our architecture is over-engineered and the customer wants to use the IoC container for it, so the issue was to get it from the controller to the call to the next proxy. I got it working, though.


I have a new problem where I'd like your help to make our project a little less poo poo. We have a web service that acts as a middle mad for communication to CRM 2011(being replaced by D365 soon). The web service is built around a classic 3 layer architecture:

API layer, which contains controllers who just has reference and calls through to
Business Logic layer, containing business managers who just has references and through to
Service Access Layer, containing the reference to CRM. Only this layer knows CRM's data structure and is responsible for mapping to DTOs which get returned all the way back to the caller.

The API layer was getting more and more controllers and methods with unique DTOs, so one of their architects got the idea of making more standard DTOs per entity, with varying degrees of fields and references. So we'd end up with something like CustomerBasic, CustomerNormal, CustomerFancy, sometimes more, sometimes less.

It's not my decision, but I thought it was a terrible idea, and I'd much rather have a DTO per use case so I can pick exactly the fields I need, but I'd rather have some sort of query language so we didn't need all those end points. The problem with this, though, is that only the SAL can know CRM's datamodel, it has to work with Swagger, and needs to work with NSwag generated proxies.

What are best practices for these types of setups? Is the entire thing trash?

Opulent Ceremony posted:

Well this sure doesn't give me a lot of confidence! We use it to encode and DASH stream our own video content.

EssOEss posted:

Boz0r posted:

It's not my decision, but I thought it was a terrible idea, and I'd much rather have a DTO per use case so I can pick exactly the fields I need, but I'd rather have some sort of query language so we didn't need all those end points. The problem with this, though, is that only the SAL can know CRM's datamodel, it has to work with Swagger, and needs to work with NSwag generated proxies.

What are best practices for these types of setups? Is the entire thing trash?

adaz posted:

The entire thing is trash yes but often times we don't get a choice in what we have to work with so here are some suggestions!

1.) Add another library project that just contains your models. If you really want to be ridiculous (and based on your architecture you might need to be) it can jsut contain the abstract classes or interfaces and let concrete implementations up to each real project. So you Controllers / SAL / Business logic all have dependencies on _it_ and it has references to nothing
2.) Change your SAL to layer to accept types of <T> and rely on reflection to populate the DTO. Your SAL becomes an ORM mapper for the uptsream projects. Your controller supplies the type it wants to your sal and your sal is relying on reflection and discovery to populate it with data from your CRM. This is tough both practically and for performance reasons but it's a common solution to this problem and meets some of your goals, is more flexible, etc
3.) Use something like automapper (NOTE: read Jimmy's best practice guidelines _and do not violate them!!_) to map your sal DTO -> Business DTO -> Controller dtos.
IN general I would avoid this because its a whole lot of object creation/destruction for absolutely (0) benefit but you might not have a choice depending on your architect.
4.) as another poster mentioned something like GraphQL or Elastic Search/Solr is basically built ot be a generic data query language around existing data sources. But that's a complete rebuild of your app which is probably not practical!

Boz0r posted:

At the moment we have a library that that generates IQueryable sets with typed entities for interaction with CRM. Would it be realistic to build a DTO library that mapped DTOs to the entities we needed and be able to send an IQueryable expression as a REST call?

adaz posted:

Yes, the hard part is then mapping REST -> IQueryable. Like REST is about resources and state and defers how to handle querying, sorting, pagination, etc basically solely up to the implementer. With that said, sure many frameworks expose REST endpoints that under the hood rely on IQueryable to sort and get the data that the REST endpoint exposes. However if the point is to expose Iqueryable<T> to the outside world from the CRM why not just let them query the CRM which surely already has a REST interface or index the data and expose it as lucene/elastic search queries which are industyr standard ish and people are more used to? Also a million clients available for

Eggnogium posted:

I have a Json.NET question:

A web API I'm calling returns some json including one attribute that is an int64. I'm taking this data and serializing it into a .NET object, but all the consumers of this object will actually want to read it as a string after converting to hex. Is there a super simple way to augment JsonConvert.DeserializeObject to just do this for me so the deserialized object looks as it should? I could have two properties, one with a setter for deserializing and one with a getter that hexifies the other, but I'd rather avoid that if possible, keep the object interface clean, and bundle this complexity with the deserialization code.

LOOK I AM A TURTLE posted:

There are probably several ways to do that. One way is to create a custom JsonConverter that does the needful. You will need to either add it to the JsonSerializerSettings that you pass to DeserializeObject, or apply it as an attribute to the class or property in question. See https://www.newtonsoft.com/json/help/html/CustomJsonConverter.htm and related articles.