What is the value in services.AddHttpClient<MyBlobClient>()

Question

I've just spent the last hour learning a ton about HttpClient (here and here and here and ...). And clearly I want to use IHttpClientFactory to get any HttpClient I need. But I have not seen a clear answer on one part.

95% of my HttpClient requests will be to my Azure BLOB storage. So same URL for all those requests. I see some examples that do something like this:

services.AddHttpClient<GitHubClient>("GitHubClient.Version9", x => { x.BaseAddress = new Uri(GitHubConstants.ApiBaseUrl); });

Where it is creating a client with a specific base address. So a couple of questions about this:

1. Is this a good idea and what is the advantage?
2. Will I get the not updating DNS issue with this? I’m guessing the IP address of my Azure BLOB storage will be changing regularly.
3. I still have to provide the URI for a specific file. So am I then just providing everything after the base? (This is a slight pain as I have these urls in the database as the full url - and they need to be that way.)

Answer 1

Is this a good idea and what is the advantage?

The whole idea behind the type client AddHttpClient<GitHubClient> is that you want to hide the http based communication from the rest of the world. The GitHubClient class can encapsulate all the following logic:

• Prepare request object
• Issue request
• Handle transient failure(s)
• Perform checks against the response
• Parse the response's body into a desired data structure

From the consumer of the GitHubClient perspective it is all hidden. (S)he can only see strongly typed methods like

• Task<List<Branch>> GetRemoteBranches()
• Task<List<PullRequest>> GetPullRequests(PullRequestState stateFlag)
• ...

For further information I would recommend the following threads:

• Theory: Typed HttpClient vs IHttpClientFactory
• Usage: Cannot get Polly retry Http calls when given exceptions are raised

Will I get the not updating DNS issue with this? I’m guessing the IP address of my Azure BLOB storage will be changing regularly.

In case of .NET Core and .NET 5+ the HttpClient instances are short living whereas the underlying HttpClientMessageHandler are pooled and reused. You don't have to care about the DNS resolve issues any more

For further information I would recommend the following threads:

• What is the lifetime of a typed HttpClient instance from IHttpClientFactory where the type that will receive it is registered as "AddScoped"?

Do I still have to provide the URI for a specific file. So am I then just providing everything after the base? (This is a slight pain as I have these urls in the database as the full url - and they need to be that way.)

Yes that's correct. If you have set up the BaseAddress then it will be always used as a prefix. So, in your requests you should only provide the relative url. Gladly you don't have to deal with the forward slashes, it will be resolved automatically. So, the bellow four code examples are issuing the same request against https://httpstat.us/200 url.

var httpClient = new HttpClient() 
   { BaseAddress = new Uri("https://httpstat.us") }; //no trailing slash
var result = await httpClient.GetAsync("200"); //no leading slash

var httpClient = new HttpClient() 
   { BaseAddress = new Uri("https://httpstat.us/") }; //trailing slash
var result = await httpClient.GetAsync("200"); //no leading slash

var httpClient = new HttpClient() 
   { BaseAddress = new Uri("https://httpstat.us") }; //no trailing slash
var result = await httpClient.GetAsync("/200"); //leading slash

var httpClient = new HttpClient() 
   { BaseAddress = new Uri("https://httpstat.us/") }; //trailing slash
var result = await httpClient.GetAsync("/200"); //leading slash