Posts

Sometimes we’re presented objects in JSON that do not directly map to a strongly typed object. Here’s a simple example: [ { "type": "Car", "wheels": 4, "trunk": true }, { "type": "Bicycle", "wheels": 2, "persons": 1 } ] The goal is to deserialize these into their respective types, Car and Bicycle, both inheriting from a shared base class Vehicle. JsonConverter JsonConverter allows you to customize how JSON is deserialized. Here’s an implementation that checks the type property and instantiates the appropriate object: ...

A common issue in API client libraries is mismatches between the JSON response and the expected strongly typed objects. I encountered this with TMDbLib and created a validator to catch these mismatches during unit testing. The Settings We can set up a JsonSerializerSettings object to: Treat missing properties as errors. Use a custom ContractResolver to enforce that all expected properties must exist in the JSON. Handle and log deserialization errors without throwing immediately. private static readonly List<ErrorEventArgs> Errors = new List<ErrorEventArgs>(); public static void Main() { JsonSerializerSettings settings = new JsonSerializerSettings { MissingMemberHandling = MissingMemberHandling.Error, ContractResolver = new FailingContractResolver(), Error = Error }; JsonSerializer serializer = JsonSerializer.Create(settings); } private static void Error(object sender, ErrorEventArgs errorEventArgs) { Errors.Add(errorEventArgs); errorEventArgs.ErrorContext.Handled = true; } This configuration makes sure that: ...

This is a short post on using Let’s Encrypt to get TLS certificates for NGINX virtual hosts. The setup assumes: Ubuntu NGINX hosting multiple domains A scheduler (e.g., cron or Rundeck) Let’s Encrypt Let’s Encrypt offers free domain-validated (DV) certificates for HTTPS and other secure services. letsencrypt.org GitHub Initial Setup I run multiple web services on VMs behind a single public IP. My NGINX proxy manages SSL termination and routing using virtual host configs. ...

Previously, I described a Bencoding Library I made to simplify working with the Bencoding structure in .NET. I also published it on Codeplex. However, I discovered a flaw—if a string contains a null-byte (as most .torrent files do), it will fail. So I made a fix, which has been uploaded to Codeplex as changeset 1102. As can be seen in the changeset, a few files have been changed. Ignoring all the test files, the changes were made to: ...

I asked around while trying to learn how to use LINQ. LINQ is essentially SQL for .NET—it allows you to query objects directly in memory. For example: List<int> test = new List<int> { 1, 3, 5 }; List<int> testnew = (from a in test where a == 5 select a).ToList(); Someone claimed LINQ was faster than using foreach for selecting items. So, I tested it. The Benchmark static List<int> testlist = new List<int>(); static void Main(string[] args) { DateTime b, a; List<double> regularTimes = new List<double>(); List<double> LINQTimes = new List<double>(); for (int x = 0; x < 1000000; x++) testlist.Add(x * 5); // Foreach loop test for (int x = 0; x < 1000; x++) { b = DateTime.Now; TestRegular(); a = DateTime.Now; regularTimes.Add((a - b).TotalMilliseconds); } Console.WriteLine("Regular Avg.: {0}", regularTimes.Average()); // LINQ test for (int x = 0; x < 1000; x++) { b = DateTime.Now; TestLINQ(); a = DateTime.Now; LINQTimes.Add((a - b).TotalMilliseconds); } Console.WriteLine("LINQ Avg.: {0}", LINQTimes.Average()); Console.ReadLine(); } private static void TestRegular() { List<int> newlist = new List<int>(); foreach (int item in testlist) if (item % 2000 == 0) newlist.Add(item); } private static void TestLINQ() { var newlist = (from a in testlist where a % 2000 == 0 select a).ToList(); } This runs each selection method 1000 times on a list of 1,000,000 items. ...