255: Rapidfire 85

01:07:43   Download

On this RapidFire episode we answer your questions on QA processes, accessibility on a gallery of pictures, gulp, Dave explains something like you’re 5, why are there bugs in software, and how to justify mobile site development to stakeholders? All this and a bit of a follow up on this episode of ShopTalk Show!

Jump to a discussion on…

  • 1:40 I’ve been struggling to make my QA process as efficient as possible.
  • 14:00 How do you label a gallery of pictures with accessibility in mind?
  • 25:00 Is there an advantage of using gulp-watchify over plain old gulp.watch?
  • 30:34 Can you explain this to me like I’m 5: How do modern MVC frameworks “connect” to the backend, or even third party APIs?
  • 44:40 Why are there bugs in our software?
  • 51:30 If you are working on a website who’s mobile traffic and conversation rate is low, how do you justify the additional time to create a mobile friendly UI to stakeholders?
  • 1:00:45 @Support


Sponsor: HelloSign 22:55

Embed documents directly on your website with just a few lines of code.

Try the API

Sponsor: CodePen PRO 43:35

CodePen PRO unlocks privacy

We’ve got some amazing new privacy features coming to CodePen Pro – nothing we can announce yet – but if privacy is a feature that’s important to you, check out CodePen Pro.

Job Mention

  • David Hicken

    Writing software does not have a small set of variables like making bread does. With bread making, you have about 3 variables: ingredients, process (how the ingredients are combined and baking time at what temperature), and environment (humidity, elevation, etc). Process can be cut and dry, environment can easily accounted for/controlled, and so the baker just needs to make sure they are getting the same consistency in the quality of their ingredients (my wife bakes, and her baked goods (bread being one of them) turn out great pretty much every time).

    Software development is more like rocket science. There are nearly an infinite number of ways it can fail, and even with some of the most minor of mistakes, it will. Developers with more experience tend to be able to write code with less bugs, but there are so many ways things could go wrong, for example: type-o’s, omission errors, unforeseen variables (different devices, different browsers, unexpected user actions), unforeseen needs to be met, changes in software that impact other parts of the software, developer having a bad day, etc. And of course the larger the code base, and the more complex the application, the more possibles there are of things that could go wrong, or not be accounted for.

  • On automatically creating alt text with an API: I think the human-supported api offered by CloudSight is up to the task. Its neither cheap (about 0.04/0.03$ per image) nor fast (about 10 seconds) but about 90% of the results I’ve seen I would consider to be decent alt text. Just now it gave me “close up of synthesizer keyboard on brown table” < thats good alt text, although the image isn't too complex.

    They also use this in an smartphone app called TapTapSee and I have a few blind friends who make good use of it. So the descriptions it comes up with are useful to non-sighted users.

    Anything i've seen from a fully AI based system would be wrong/misleading to often. E.g. Facebook uses AI to offer image information to screen readers in their mobile apps. Its not a fully formed description, just a few keywords. It can still be very useful in the Facebook context. If you have people already tagged in the image and then a few keywords like 'group', 'smiling', 'trophy' you know enough. You also have a good amount of context around the image with the post text and the comments. For completely arbitrary source images, it wouldn't be so useful.

    I've been meaning to write a wordpress plugin with that API that automatically pre-fills alt text for uploaded images. The free 500 images/month plan would probably suffice for a lot of sites.