DEV Community: sige

What can European SaaS companies do after EU-US Privacy Shield invalidation?

sige — Thu, 24 Dec 2020 14:05:15 +0000

There is not much hype around this topic however, I believe it is very important for EU businesses today.

The European Court of Justice declared the EU-US Privacy Shield to be ineffective due to the FISA 702 regulations in the United States. This basically means that US authorities can execute targeted surveillance on non-US citizens and companies. They can collect, keep, and use this data for investigation. This regulation does not affect US citizens and businesses.

Our company ConfigCat.com serves feature flags and is the sub-contractor of other companies as part of the supplier chain. Some of our European customers started to worry about if their data is safe with their sub-contractors and how does this affect GDPR.

Being a Europen SaaS company with a global user base we looked into how we can handle this quickly with minimal user impact both technically and legally.

If you are facing or faced similar issues, I'd like to hear your thoughts about it. Do you think we are going to face an increasing number of privacy issues in the future due?

Here is what we did:

Implemented a Data Governance feature in ConfigCat, so customers can decide if they want their data distributed globally or only in the EU based CDN nodes.
Reviewed all our own sub-contractors swapping the US ones who fall under the FISA 702 regulations.

I'm personally having a bit of anxiety about where this is going. Ideally, every entity should be able to use any service no matter where they are located without being worried about the safety of their data. Maybe I'm just too idealistic.

Our blog post on the topic: https://configcat.com/blog/2020/11/28/eu-us-privacy-shield

Releasing new features to a small audience first using feature flags

sige — Thu, 14 Nov 2019 14:10:12 +0000

Do you have a product with a consistent user base and you are releasing new features? Then you need to decide whether it’s enough to just test these features in- house. This decision is always tricky, because sometimes it really is useful to receive feedback from a wider audience without the risk of imposing the new feature on your more important or so-called “sensitive users” (major companies that were pleased with your facilities and services up to the present, and you cannot risk the collaboration with them).

If you are facing this dilemma, ConfigCat‘s feature flags for targeting might provide you with a solution.

Let’s say you have a product like Instagram, and the company decided to add new poll-boxes, quiz types and options for embedding links in the stories while removing some of the older ones.

Using feature flags you will have the opportunity to enable certain features to certain group of users. Let’s say you would like to enable a feature only for the users within your company or only to a small percentage of your users (maybe in one country only) before releasing it to the entire world. In addition, you might want to exclude big companies, like H&M or Victoria’s Secret until you have received the favourable feedback from your employees, “friendly users” and a small percentage of app-users or even users from a selected area (country).

Even though in this article it’s exemplified with ConfigCat, targeting works just fine with other feature flag tools as well — for example with Unleash-hosted. The process of targeting special users goes down in the following way:

First step: you need to enable the feature:

Log in to access the Management Console
Go to Feature Flags & Settings
Select TARGET SPECIFIC USERS after clicking the actions icon

After you turned on the feature flag, you will be able to target specific users.

How does the targeting rule work?

By adding a rule you can specify a group of your users whom a new feature will be enabled or a different value will be served. A rule consists of an Attribute of a user in your application, a Comparison value (e.g. a list of email addresses) and a Comparator. ConfigCat evaluates the targeting rule every time your application requires and decides what value to serve.

A property of the user is considered as an attribute, and it can vary between e-mail address, country, and identifier or custom (you can add highly specific attributes such as eye colour).

The comparison value is any string, number or comma separated list to be compared with the selected attribute using the comparator. It can be a list of e-mail addresses, companies or users in a specific country.

The comparator has four different setting options and it marks the selection of the users with a specific attribute from the list (comparison value).

The term ‘served value’ mentioned before refers to the exact value that will be served to the users that fall into that fraction. Depending on the kind of your setting this could be:

In our example the On/Off Toggle is relevant because that is the usual state of a feature flag, but ConfigCat offers a range of different setting kinds and types.

You actually have multiple options in choosing the served value of the feature flag, for example, it can be the On/Off Toggle presented above (note: here the number of options must be 2), but if you decide to use Text and/or Number the maximum number of your options entirely depends on your subscription plan. You can add/remove options by clicking the Actions icon.

Once you’ve enabled your feature for the list of your employees for instance, you can add multiple targeting rules and ordering by clicking on the Actions icon; thus you can create more complex rule sets.

Rule sets are evaluated one by one, from top to bottom direction.
Change the order of targeting rules by drag n’ drop.

Example

Enable a feature only to users within your company except the PR team (Connor and Natasha) by adding two targeting rules:

To all other uses serve: OFF

This value will be served as a fall-back if none of the above rules apply or a User Object was not passed to the ConfigCat correctly within your application.

You can also target a percentage of users. If you enable a new feature on a platform like Instagram, you might want to be granted public feedback by including a randomly selected fraction of your users whom a feature will be enabled, or a different value will be served. The sum of all % values must be equal to 100.

Select TARGET % OF USERS after clicking the actions icon.

To sum up, targeting can prove to be great help in experimenting with new features while offering you and your company a risk-free path to gather feedback from the targeted users. Feature flags provide you with a safe environment with the option of changing your mind.

How do you recruit for user tests?

sige — Mon, 09 Sep 2019 10:22:50 +0000

I feel I got lost in the maze of Google searches filled with advertisers and self marketing articles. Seems quite hard to find authentic information about user testing.

Where do you guys usually recruit subjects for user tests?

I'd like to test my service on 3-5 software developers with a few years of experience. I believe my needs aren't that extraordinary.

Do you guys offer money or a subscription to your service in return?

Thanks for the help. I'm a little clueless here.

Playing around on CodePen I have created a simple HTML page with dynamically updating content based on feature flags

sige — Sun, 08 Sep 2019 23:51:17 +0000

This Pen demonstrates how to update a page content dynamically using ConfigCat feature flags.

Complete guide to build a Slack Chatbot in 7 minutes and host it for free

sige — Fri, 06 Sep 2019 22:27:35 +0000

Based on GitHub's Hubot. Connected to Slack. Deployed to Heroku.

I have a thing for robots and wanted to build a chatbot as an experiment that calls my service's (ConfigCat.com) \health endpoint and returns if everything is okay. It took me quite a long time to go trough the possible frameworks and docs, read all the outdated guides to find the quickest and cheapest way. I feel like it might worth sharing.

The tools needed:

Node.js installed
Slack workspace
Free Heroku account
Preferred code editor

Adding Hubot integration to your Slack Workspace

Find Hubot in Slack App Directory

Connect to Workspace

Take a note of the API Token, you'll need it later

Running Hubot on local machine

Install Yeoman and Hubot generator

npm install -g yo generator-hubot

Scaffold a Hubot project

mkdir catbot
cd catbot
yo hubot --adapter=slack

Start Hubot using API Token

HUBOT_SLACK_TOKEN=xoxb-271695489427-739714865891-Z5gPPiuTORKDFO4QvqKe1B9y ./bin/hubot --adapter slack

Open Slack and start a conversation

The chatbot should be available under Apps.

Test with `help` command

Recognizing `health` command

And make a HTTP GET call to ConfigCat's /health endpoint and reply back the results via Slack.

I created a configcat.coffee under /scripts folder with the following code:

module.exports = (robot) ->
    robot.hear /health/i, (reply) ->
        robot.http("https://api.configcat.com/api/v1/health")
        .get() (err, res, body) ->
            reply.send body

See the complete source code on GitHub.

I used CoffeeScript because I like to experiment and it feels fancy. But you can use JavaScript as well.

Test if the `health` check works

Deploying to Heroku

Check Node.js version

node --version

Open package.json and check Node.js version. In my case the generated package.json was "node": "0.10.x" so I changed it to:

"engines": {
  "node": "10.16"
}

Git commit

Make sure you’ve created a git repository, and that your work is committed.

git init
git add .
git commit -m "Initial commit"

Heroku CLI

Install the Heroku CLI. then log in.

heroku login

Create app.

heroku create

Set environment variable for API Token.

heroku config:set HUBOT_SLACK_TOKEN=xoxb-271695489427-739714865891-Z5gPPiuTORKDFO4QvqKe1B9yt --app=arcane-dusk-29327

Git push

git push heroku master

Avoiding sleep mode

Since I'm using Heroku's free plan, the app will eventually go to sleep mode. To avoid that add hubot-eroku-keepalive script to your chatbot.