Writing KelpFT

2024-08-13

I have been using fitotrack for tracking my workouts since.. pretty much when I started cycling, about three years ago, and it has been... fine-ish. I am happy with the core functionality but i have three problems:

And so, considering that I have been itching to program something once again after writing [My first freewheeling application], I have a new mobile project in mind with this feature set:

As for which tooling I should use for this.. I think i'll have to lean on flutter for this one. Although I wouldn't mind trying out jetpack compose.

I must say that trying to use LOVE for something like this will prove too difficult and time consuming, in my opinion.

2024-08-20

I've finally decided to commit a little bit harder to this as I have run out of other things to do at the moment.

I'm calling it 'KelpFT', it is going to be written in flutter.

I'm dividing my work into two rough fields. One for technical solutions and another for graphical screens.

The distinction is drawn there because these two sets are distinct in how i think about them. One set requires me to solve hard, practical puzzles essentially, while another requires less logical thought but more attention paid to how things look visually.

Technical solutions needed:

Screens needed:

2024-08-29

And here we are, nine days later. I've been putting in a few hours from time to time into the app and i've gotten to the point where I am ready to begin basic field testing.

So, to put it simply, I ran into three major technical hurdles:

First off, It is not feasable to have a file-based data store for reading and writing and manipulating workouts. In other words, using a .gpx library (which I was fortunate to find) to read each and every file upon app start is not feasable. The reason being that there is simply too much overhead. In short, I ended up using the library to deserialize the files and noticed that it was taking far longer than I wish it did (think 3-5 seconds on an average phone). I was chalking it up to Flutter/Dart/the GPX library I used being too slow but honestly, .gpx files contain huge amounts of text. For reference, one of my four-hour-long endurance workouts chalk up to be a litte over 41 thousand lines of XML, and at this point i have about 50 of these files of varying length. No wonder it was taking too long. This essentialy left me with two choices:

Second, google, as a company, is really, really getting in my way of just letting me write whatever program I want. The direction that android has been heading towards as of late has been abysmal. For one, they have been pushing a new geolocation API for a while now, that, for whatever reason, requires the google play services installed to use the local gps radio of your device. One of the flutter libraries that I was using in an attempt to get GPS geolocation working was relying said API. I had to learn that fun fact the hard way, since I am running LineageOS.

Luckily there is another flutter library for getting your location that instead relies on LocationManager for it's android implementation, which does not require any google services installed.

And last, My home, or rather, my room, is unable to receive GPS signals! I cannot test the core function of this application from the confines of my house. I'll need to conduct some more field testing.

What I've progressed on so far

Well, essentially, i've been putitng one to two hours every day into the app, and so far, I have:

What I need to work on going forward

In terms of relevance:

2024-09-01

Progress Update:

I have also a few other things i'd like to address. For one, I am thinking of using something other than sqlite. At the end of the day, i was either going to go with an entirely filebased approach or a database-based approach, and if i'm picking a database, i might as well pick the one that is best suited for my needs and still maintain a common export format (gpx).

On top of that, since It makes for a much easier dev experience (you store objects directly, instead of converting them into SQL statements), I want to try using ObjectBox as my database, since I am not interested in being able to transfer data directly to and from the application any more, as that would be handled by .gpx import/export (I'll need a way to import/export in bulk).

2024-09-03

Progress Update:

Hm, what do I need to do to deem this project complete for the time being?

2024-09-07

Progress Update:

2024-09-09

It was somewhere at this point I decided to finally set a deadline for myself. Marking the 20th of this month as the day I declare the app 'done', which means that I gave myself a month to go from first line of code to completed project.

2024-09-11

It's the 11th as I am writing this and I am proud to say that I have been able to steadily make progress on the app for the last two weeks or so. At this point, 90% of the app's UI is fleshed out, I just need to draw a few additional icons, and make the last screen of the app, which just contains settings.

I am currently facing a dilemma when it comes to picking out the UI approach to delete workout entries. I am stuck between a plain 'longpress-to-delete' and 'swipe-to-delete'. After thinking about it for a bit, i'm leaning towards 'swipe-to-delete' because i've never done something like it before, and I want to add the technique to my arsenal.

2024-09-18

Unfortunately I ended up missing out on the last couple of days in terms of coding work due to spending one day celebrating a birthday and spending the next day after that doing yard work. What's even more angering is that I have also given myself a leg injury from said yard work, which has impaired my ability to walk/cycle for the next couple of days. I can write code for this app but I can't test it as well as i would wish.

So, it is the 18th as I am writing this. I have given myself 48 hours to finish this application, so far, I think i won't be missing too much.. The application, as it stands, still needs a fair bit of polish, but, as of now, what I have is:

Were I to think in terms of delivering an MVP, i would only really need a full backup functionality at this point, which can be achieved with a full .zip export of all of my workouts, and i'd also like some assurance in terms of the app not getting killed after a few hours of working out.

2024-09-20

Well, i've managed to achieve most of what I set out to do, import/export works, and I am forced to move on to other projects. The app is.. mostly there. I still need to re-reimplement the tracking API because it does not work reliably enough.

The source of the app can be found here