Category: Software Development

As the inception of a new project takes place, it is crucial to establish whether the set deadline is feasible. This constant evaluation serves a dual purpose: it allows me to decide if the teams need to extend their working hours, or if we require additional resources from senior management. The keyword here is “early” — incorporating extra resources at a later stage in the project might not significantly influence the end date. Hence, leading indicators are critical early on in a project.

A simple, yet efficient, way to stay ahead is by regularly tracking the projected completion date. This process commences with a preliminary assessment of the duration needed for the team to wrap up the development. The next step involves dividing the outstanding work by the team’s velocity, measured by the quantum of work completed in a day/week/iteration. This gives us a broad idea of the anticipated date for development completion.

Though this calculation appears straightforward, real-life projects entail subtleties. One, team velocity varies from one iteration to another, influenced by factors like holidays, leaves of absence, changes in team composition due to joining or departures, etc. While dealing with this, we can either resort to the average or moving average of team velocity or rely on experience to set a benchmark.

Two, the scope of a software project tends to expand over time as we unearth new tasks or requirements during development. Assessing the remaining quantum of work at any given point in the development cycle can pose a challenge. We can either total all the active work items (requirements, user stories, and tasks) or estimate the final size based on the initial size and deduct the completed work from the estimated final size. The choice of the method depends on how work items are managed and updated.

By doing so, we can gauge the days needed to complete the development. Adding this to the current day gives us the projected development completion date. If the projected date falls later than anticipated, it’s time to spring into action. This may entail extending the team’s working hours or strategically adding more resources early on to meet the project deadline!Regenerate response

 276 total views

I have always been skeptical of using the low-code or no-code platforms. In the past two weeks, two use cases came up.

AppSheet

First, our company needs a time tracking tool. Even though my company plans to use Clockify for time tracking, I thought I’d try one of the no-code platform: AppSheet.

pro: I was able to develop and deploy a basic time tracking app on my phone in about 4 hours. It reads an Excel stored on my company’s network and build a starter app in a few minutes: one view to show all the time entries in the Excel, one view to view the details of an entry and another view to edit/add an entry. I spent the rest of the few hours to customize the app by changing settings of views or data properties(enum, editable, etc.)

con: I would like to make it a bit fancier: when user starts working on a project, she hits the start button to time the project. When she is done, she simply hits the stop button. This would be straightforward to do on any mobile app development framework(Ionic, Java/Android, Objective-C/iPhone), I found it very difficult to do in AppSheet.

PowerApps

The driver to try Microsoft’s Power Apps is to view DevOps work items on my mobile phone. What I want it to do is let me select a DevOps query to run, displays the work items and detail of a selected work item. I was able to complete majority of the development on a volleyball tournament over the President Day’s weekends.

pro: PowerApps is more flexible. You write code like Visual Basic. You can add a canvas. Add controls to the canvas. Modify a control’s properties to change its user interface. Or bind an action(e.g. OnSelect) to a function call(e.g. getting data or navigating to a different canvas), like what you do in Excel(e.g. sum() or average())

con: the action only allows one function call. You can’t write more than one line. It is by design so that PowerApps developer don’t need to write(and debug) a lot of code.

Conclusion

By no means, this is a thorough review of both platforms. But one take away is, like all platforms, they offer convenience by taking away some complexity of coding. However, this also takes away flexibility and power of coding. IMO, PowerApps strikes good balance of convenience and flexibility. They are great to develop simple applications.

I still don’t believe no-code or low-code should be used for enterprise application development, just because of the complexity and massive size of work. I don’t see how tens or hundreds of developers can develop the software collaboratively without overwriting each other’s work. Debugging is also problematic.

 271 total views

This is the best story about MVP(Minimum Viable Product). A good story for all the product managers or whoever manages scope of product releases.

If you don’t have time, watch the video around 6:30.

In short, Nvidia was running out of money. The first(?) chip RIVA 128 supports only 8 modes of the 32 blend modes specified in DirectX. Nvidia went out to all the game developers to convince them to not use the other 24 modes. “8 modes is all you need”. “if you want to make an explosion, you want to make transparent, could you do it this way? use that mode. just don’t use the other 24 modes.”

 272 total views

Somehow, the software developers I’ve met were very difficult to convince. Even as simple as using tabs in code for readability. I, at times, were difficult, too.

I found a script one day. It worked most of the time, if not every time.

I asked engineers to have the product send out email notifications when the product runs into an issue. As usual, they passively refused to do so. One day, at 6 pm, I found R sitting in his cube. I asked “R, shouldn’t you go home now?” He told me he was waiting for a long-running ETL script to finish. He needed to babysit the script so that if there were any errors, he could take care of it as soon as possible. I said that you should implement the email notification so that you can go home and only hop on then computer only if you got an alert.

They key is to tie the desired behavior to their benefits. Since then, I have been using the same trick.

If one is going to have a long vacation, I would ask him/her to document things like how to support the product, troubleshooting steps, etc. so that I don’t have to call him/her to be log in to fix the issue while on vacation.

 283 total views