I’ve written before about how it’s better to write straightforward code than trying to be clever and pre-optimize. Another example of where this rule is true is the cost of iteration between different styles of array loops in JavaScript when executed within NodeJS.
Operator precedence determines the order Operator precedence determines the order and binary operators are executed. In many situations, knowing basic mathematical operator precedence (for +,-,*, etc) is enough. In other situations not knowing the specifics of language can lead to hard to find errors. Quickly, in JavaScript what does 'the boogeyman ' + false ? 'is scary' : 'is not scary' evaluate to?
Recently I started writing a new web application which uses OData V4 as the protocol for passing data to back and forth between the JavaScript client and ASP.Net server. The client is written using AngularJS which doesn’t have any built-in facilities for working in OData APIs. In order to avoid writing my own OData message handler I used the Apache Olingo library to handle the grunt work of sending requests and receiving responses. I wanted to encapsulate the Olingo library in an AngularJS service in order to make using it easier, the following post details how this was done.
Martin Folwer’s blog and website is a common reference for modern design pattern so we will start with his glossary definition for the repository pattern. The linked article contains more detail, but in short a ‘repository’ acts as an data store layer on top of the storage system and provides (more) object oriented methods for accessing said system. Unfortunately, something I have also seen is the over application of this pattern in combination with RDBMS ORMs leading to contorted code with unnecessary interfaces and classes. This post covers what I have seen in some projects in order to provide some of the code smells to avoid.
How to represent a blank user input is a reasonably innocent design question. On a couple occasions though I have been bitten by systems that have been designed without considering that question seriously. Maybe it’s a clear choice to some, but today’s post makes an argument that in representing an empty input as a null is better than using a empty or blank string.
IT consulting is an odd place to be when it comes to software engineering practices. We often end up writing software for business groups that have fixed budgets and more importantly fixed deadlines. We also try to follow an agile methodology for software development that roughly follows scrum (warning: PDF), but with defined roles for a project manager and a development lead. Today’s post discusses some of the difficulties seen in using Agile to deliver business software and how we can mitigate those difficulties, basically things that worked and things that didn’t.
Being an IT consultant has occasionally had me working in unusual or cramped quarters at client offices, the result of being hired help instead of a full time employee. One type of psychical situation I have ended up working in a few times is something we called “the project war room.”
Something introduced fairly early in programming courses is the concept of an interface. In practice interfaces help to define the boundaries between components of a system, define the behavior of underlying implementors of the interface, and make it easier to switch implementations of said behavior down the road. Today’s post offers a few thoughts on where interfaces should and should not be used.
Today we will look at whether it’s worth developer time to pre-size ArrayList objects in Java application code. We will take a similar approach to the one taken in a prior post on how long it takes to handle an exception in Java.
Today post is exceptional, we’ll take a brief look at the time cost of throwing and re-throwing exceptions and put that time cost in context. To be specific, we’ll look at the timing information for a set of exceptions which might be found in a typical 3-tier business application. In our example the exceptions are thrown and caught in hierarchical order in order to promote separation between the tiers of the application.