Trntxt Autumn Update

I’ve made a couple of updates to trntxt since July.

One of the things I noticed during my user tests was that people typed ‘bristol’ when they wanted information about Bristol Temple Meads, Bristol’s main station. Unfortunately ‘bristol’ is a closer match to the smaller Bristol Parkway station because it comes above Temple Meads alphabetically. Before August, there wasn’t a way to get the user to correct their search, but trntxt now has search suggestions!

The feature was implemented while being moderately inebriated by a single bottle of Henry Weston’s at the Meanbee hack night in Bath.

Now if you go to http://trntxt.uk/bathspa/bristol it will ask you if you meant Bristol Temple Meads instead of Bristol Parkway.

Just don’t ask for ‘London’.

20151002_205404000_iOS

The other new thing is journey times. Trntxt now shows you how long each journey takes, so you can identify the faster trains more easily. The number is calculated with preference to the estimated timings, so if a train is delayed, the time will adjust accordingly.

20151002_210611000_iOS

Next logical step – sort by arrival time instead 😉

I’m also looking to make trntxt’s suggestions cleverer. I’ve started collecting data about the devices that use the site and the inputs people give. I’m hoping to expand that to also collect and store data from the National Rail Enquiries API. Don’t worry, the information collected is not personally identifiable (unless your user agent string is “HI MY NAME IS DAN”).

Hopefully, after some more practice with MongoDB I’ll be able to draw conclusions predicting the stations people mean based on their input. For example if someone types /bristol/london, they probably mean London Paddington, whereas /southampton/london should take the user to London Waterloo instead.

JSOxford NodeBots Day

“The problem with hardware is that there is no version control” – Marcus Noble

JSOxford’s NodeBots day marks one year since I first used JavaScript (ish). At last year’s event I took an Arduino and used the johnny-five library to control it using Node. It was fun to make, but wasn’t really that impressive.

This time round, JSOxford had a bunch of Espruino Pico boards and plenty of hardware to hack around with, including continuous servos and wheels.

https://twitter.com/misssarahwarren/status/624871571300200448

Continue reading “JSOxford NodeBots Day”

Trntxt bumper update

It’s been a while since I’ve posted an update on trntxt, and in that time lots of things have changed. Most of those things haven’t changed the site’s appearance at all, but I can assure you that I’ve been very busy with it over the past few weeks.

Service messages and bus times

Trntxt now shows you service messages from National Rail and timings for any rail replacement bus services. Days and days of work condensed into a single sentence…

nrcc

In order to make those changes simpler to implement, I had to make quite a few changes in the background. In doing so I learned more about the Jade templating engine and created a Mocha test framework with the help of Thom Wright at one of the Hack Nights run by CodeHub Bristol.

Little changes

Better station search

When typing a station name in the address bar, a more complex search takes place. Beforehand in order to match a station, the input had to match a portion of the station name exactly. Take for example, the station ‘Heathrow Airport Terminals 1, 2 and 3’. Originally, if you searched for ‘terminal1’, it wouldn’t produce a match. Trntxt removes all spaces and punctuation from the station name, but even then ‘terminal1’ doesn’t appear exactly in the station name – you’d need to have ‘terminals1’ which isn’t immediately obvious. Now it checks the station name to see if all the characters are there in the right order. You can type ‘lonwat’ for London Waterloo and it will return the match.

While writing the improvements to the search function, I made sure that the changes didn’t break anything else by running the tests. The test framework essentially contains a list of inputs along with their expected outputs. When I wanted to write a new feature or fix a bug, I’d put in some failing cases with the result it should return, then I carry on coding until the tests pass.

By running the tests automatically each time I save the code, I can be confident that the new code doesn’t negatively impact any of the past test cases because they are all checked every time.

tests

Appearance improvements

appearance

I decided that while still being minimal, I could improve the appearance of trntxt’s output. The 4 dashes I originally had to separate items have been replaced by a line; I’ve reduced the wordiness of each item description and the CSS fits inside a tweet (just because)!

#TweetYoCss

Preventing unnecessary app icon downloads

While investigating the network traffic to trntxt from different devices, I noticed that Android devices wanted to download the high resolution app icons for Apple devices, which is a bit of a waste when that icon is never shown. I now run a check to see what device is accessing the site, and only send links to appropriate icons for that device. It’s fairly unnecessary and especially not worth the effort to do it, but I did it anyway. Turns out it’s easy enough to modify the content of Jade documents based on various parameters, plus I was annoyed that all the icon tags almost doubled the size of the response (see below). I probably shaved nanoseconds off the page load time.

appicons

More to come

I’m not finished yet! There’s always something else I want to add to trntxt. Search suggestions coming soon!

suggestionMockup

Why trntxt beats your train times app

I’ve put a lot of work into making the trntxt experience as efficient as possible. Even though it was designed for old phones with a slow connection, it works well on modern smartphones, and in my opinion, beats any train times app out there. Of course I think it’s better (I did make it after all), but I’m going to try and explain why it’s better.

I once learned that when someone looks for information, it is best to have that information displayed after as few gestures as possible[citation needed]. My experiment compares the use of the National Rail Enquiries (NRE) app with trntxt, on both initial setup and subsequent use. I’m going to dig out my old iPod Touch for this because my phone doesn’t do screenshots and nobody else has Windows Phone 7. At the start of the experiment, we are going to assume that you know the name of the app but it isn’t installed, and you know the URL of trntxt but have never visited it before.

Continue reading “Why trntxt beats your train times app”

Let’s talk about mobile data

In this post I’m going to talk about what I’ve done (and what I’ve not done) with trntxt in order to get the fastest page load times possible. I’ve split this up into two sections. The first part might be interesting to most of you, whereas the second part goes into technical detail that will look like complete gobbledygook if you haven’t programmed using Node before.

You have been warned.

Continue reading “Let’s talk about mobile data”

Train Times Made Simple With trntxt.uk

TL;DR: I made trntxt.uk, a data-friendly train times website.

Over the past few weeks, I’ve meed putting a little bit of my free time into developing a train times website.

The idea came to me after a JSOxford talk by Tom Lane about open rail data. It turns out that the data you see on train station departure boards is all accessible, so long as you register for a free API key.

I decided that the reason why the train times app on my phone wasn’t working was because the data was taking too long to arrive over a 2G connection, so I thought I would try and speed things up a bit. I’d make a website without using anything unnecessary, and in doing so I’d have a website that works well on whatever device you’re using.

trntxt.uk

I call it “Train Text”. It works like this. Let’s say you’re at London Paddington and you want to find out when the next train to Bath Spa is.

You would point your phone’s browser at trntxt.uk/paddington/bathspa. You don’t need to use the full name of the station – just use part of the name and trntxt will probably find it. Just ignore spaces or any other punctuation. If you know the 3-letter station codes, you can use those too and save some typing time in the process (trntxt.uk/pad/bth). trntxt makes these station codes visible to help you remember them.

The website was designed so that even my old Nokia 6303 can access it correctly on a slow 2G connection.

image

I’ve stripped out everything unnecessary so that you can get the fastest responses possible (more on that in another post). That’s why it doesn’t look very good 😛

There are, however, some limitations. One of them being it can only handle direct routes. If you’re traveling from Swindon to Oxford, you need to change at Didcot. You’ll need to make a request for Swindon to Didcot, then another from Didcot to Oxford.

It doesn’t do arrival times yet, but I’ll get them in soon!

What’s more annoying is that after I made the site, I realised that it seems to work everywhere apart from Bath Spa station.

image

Party in the City

Last night, Bath hosted its annual ‘Party in the City‘ – a night of music and arts in multiple venues across the city. Even if you stayed for 10 seconds in each event, you probably still wouldn’t be able to see everything.

I was supposed to be singing with THE Bath Chorus* in the Abbey, but I lost my voice 😦

Instead I wandered around and watched some of the other acts and took some photos.

There was a samba band in the Circus

IMG_5501 (1280x853)

Continue reading “Party in the City”

Realtime hacking with JSOxford

Yesterday I went to a hack day run by JSOxford. The theme was ‘realtime’, i.e. using Web technologies to update a site automatically from a data source.

Since I had no experience with realtime technologies before I came, I didn’t want to make anything too ambitious! I just wanted to learn the basics so that I could make something useful in the future.

Continue reading “Realtime hacking with JSOxford”

Breaking News

I interrupt this extended period of non-blogging-ness to bring you news that isn’t interesting to anyone.

I’m not exactly what you would call a health freak. Since moving back to Bath to start my job I’ve been thinking “I should join a badminton club” or “I should join the gym”, but in 4 months that still hasn’t happened. As far as exercise goes, I walk to work every day – half an hour there, half an hour back – and that’s pretty much it.

I hate running. Sometimes I’ll try jogging to the Co-op (it’s literally just at the end of my road) but I always get so out of breath it put me off trying anything more strenuous.

But yesterday, the oddest thing happened. I woke up and thought “I should go for a run”. But it was a bit rainy that morning so I didn’t bother.

Then this morning, the weather was calm and bright, and I saw this tweet.

That tweet actually inspired me to put on my sports kit and (for the first time ever) leave the house with the sole intent of arriving back a few minutes later with an increased heart rate.

I ACTUALLY WENT FOR A RUN

And I did what the tweet said – I went on an incredibly short 1km run, but not quite as short as the one in Run Fatboy Run.

Turns out 1km isn’t very far at all, but I was still breathless at the end. Not quite a half marathon just yet, but maybe one day I’ll enter one!