Giant Devcast: A Tale of Two Dans

Hey. That was a trick. I thought it was a new show called double dan.

Nope! Try again!

GangGang DanDan

Goddammit Dans.

You've got questions, we've got Double Dans.

Thanks for this folks. I'm an engineer running a big ol sales API in C#/asp.net core and i sent THIS podcast to IT ops to get GB removed from the filters.

'Please update filter from 'gaming' to 'software related'. See? it says REDIS. IM LEARNING THINGS'

@wcarle I saw Rasmus (the creator of php) speak at Etsy 3?ish years ago on PHP 7. Some real fun takeaways:

1. While working on PHP 6 he doubled down on 'i don't think php should be used as much as it should and it should be dumb and simple and use Java for big applications' but the rest of the working group told him 'look. what you want and what the world is doing are very much NOT the same. get your shit together' and thus 6 was scrapped and 7 was born. So no, it does not exist! It was basically just 5+

2. They were actively targeting HipHop from facebook for perf gains. But to get those perf gains WITHOUT actually doing the hip hop virtual machine pre-compile. They also worked directly with wordpress to figure out paths to optimize, which i assume help with a site like giantbomb in delivering content.

Whats your deploy process like? Yall on that wonderful dark deploy CICD mess yet? I'm working on that for my apps, but we're pretty legacy. I've also got active business hours to support so my life is still 6am deploys.

Maybe not the best place to write this but hey ..devs might be listening ..worth a try..

Great work on the new site guys ..no seriously love it..

So Uhm small question.. can anyone explain to me why the video carousels are inverted on the front-page.. and if this is intentional could users get like a "invert (carousel) axis" options in the user settings. Having the carousel right button going backwards in time hurts my brain .. seriously.

(front page carousel direction is also different than on feed carousels like podcast themselves )

Now ill let you guys do what do you ..

Cheers

@danauer, how do you manage your design libraries? I've used Sketch and Figma and I know you've touched on both a little bit.

Another question would be what are your most used plugins for Sketch?

Thanks again, guys! Looking forward to the next episode

This was my first time listening to the Devcast and it was really interesting hearing about the server setup/architecture powering Giant Bomb (20 bare metal servers if I understood correctly is very many servers to me) and Gamespot. Funny story about the 404 search bill as well :)

I have a question that is more related to process and management than technical stuff so feel free to skip it if it isn't relevant.

You mentioned how upgrading to PHP 7 took several months but I thought I heard you talk about sprints (apologies if you have described this in a previous podcast) as well so my question is:

Do you split up long term tasks into several chunks that might fit into sprints (maybe with some deliverable for each one) or do you reserve part of each sprint e.g. 20% of the time to work on a long term project resuming the work where you left it?

Personally I think agile methodologies have great principles that often work well but sometimes it's as if they get hijacked as arguments for postponing necessary redesigns or migrating to new frameworks.

DAN DAN kokoro hikarete ku

Sono mabushii egao ni

Hatenai yami kara tobidaso　HOLD MY HAAAAAAND

How do you guys handle documentation for the site's functions? Is there a master list of noes, functions, styles, etc. that people can refer back to when building something new or is it all jsut comments in code and experience?

I did not quite expect to hear about machine learning on this cast! I do work in the analytics field so suddenly hearing someone on Amazon ML made me suddenly pay a lot more close attention. It sounds like you're training a simple classification algorithm! Hopefully someone will be on the next pod who can answer my questions about the methodology in place.

I'm assuming you're doing this fully automated while providing trueness/falsehood to each kind of spam/non-spam and then letting it do its thing. But there's quite a few methods one can employ to train using this kind of data set, so I'm curious about which exact algorithm you're using? Are you using SageMaker or some other marketplace algorithm, or even your own? I would love to hear some in-depth chatter about what's making that there classifier tick, though I fully understand if you don't want to potentially teach spammers how to avoid your system.

Speaking of analytics, is there a chance you could get someone on the pod from that side of the site? I realise it's not as "dev" as writing PHP or making a REST API or defining a design for a CMS, but it's still really cool stuff if there's someone in the back there who does cool stuff in Python or R or something like that who informs decisions in the development and/or content production process.

I just got a laptop with a high res screen (it's like Retina but it's a Chromebook), and I noticed that some of the thumbnails on Giant Bomb look better than others. On the homepage, for example, the three recent videos use thumbnails that are 640x360 -- like this:

https://static.giantbomb.com/uploads/screen_large/34/343190/3079425-cp_thisistherun_02062019.jpeg

But carousels use lower resolution thumbnails that are 320x180 -- like this:

https://static.giantbomb.com/uploads/screen_small/34/343190/3079425-cp_thisistherun_02062019.jpeg

If the carousels used the 640x360 images, they'd look a lot better on retina-like displays.