Peak AWS? Do we still need AWS?

I am reacting to the following post, because it matches my recent experience with AWS: AWS Elasticsearch: a fundamentally-flawed offering – spun.io

That’s my overall experience with AWS as well. They used to be quite reliable and have good support, but I have found that over time, the overall service is not as good or as essential as before.

If you use services that are not simple base EC2 instances, the costs are very high (like for example DynamoDB) and certainly too expensive to build a platform leveraging those added-value components at scale.

As an Open Source developer, I am also annoyed by their recent new offering that consist in taking Open Source software and build a SaaS offering on top of it. The main problem is that they do so without working with the main stack developers. They do not contribute back their code change either (to my knowledge). This end up with the kind of offering like AWS Elastic Search that is both hurting Elastic Search company, Elastic Search project and end up providing a bad experience to their customers.

To be fair, this is a more general question for the governance of Open Source projects, though, as you cannot expect cloud providers to pay fees at this scale. Passed a certain point in popularity, Open Source projects likely need to be led by a mixed team of contributors from a larger range of companies. The same question would arrise on Redis or MongoDB for example.

Anyway, this race to expand AWS offering through integration of Open Source components as a service or to provide higher value APIs seems to be at the expense of quality and customer service.

There is now plenty of decent alternatives for the base hosting part. So, I am wondering if we have passed the point where we don’t need AWS anymore.

Do you have the same experience?

During the next two days, I will be attending @frenchkitconf, an iOS / MacOS Developer Conference in Paris. If you attend as well, come say hello!

Sign in with Apple feature is finally fully compliant with OpenID Connect

The Sign In with Apple feature has been launched with iOS 13. It allows to sign-in to web sites supporting Open ID Connect, but also to create account and sign in to iOS applications. It was well received when announced at Apple WorldWide Developer Conference (WWDC) in June 2019 because the feature is design to protect user privacy. It even offers an option to avoid sharing your real email address with the compliant service, by letting Apple use a random email address and relay the messages to your real email address.

However, the feature was also criticized because it was mostly based on OpenID Connect but with a few differences and variations, making it noncompliant with OpenID.

Apple worked with the OpenID Foundation to solve the compliance issue and finally, the OpenID Foundation wrote an open letter to Apple stating they are happy they made all the changes to be compliant with OpenID specifications.

This is good news for the Open Web.

Spotted via HackerNews

SwiftUI is starting to work on iPad Swift Playground app.

I tried it tonight, along with Dark Mode support. It is still early stage but pretty nice to be able to experiment with SwiftUI on the go :)

Microsoft Dual Screen Upcoming Devices

Upcoming dual screen Surface devices look interesting. I am curious to see if the actual device will live up to our expectation.

It is also good to see Microsoft embracing Android to replace Windows on the smaller form factor, while I wonder why it would not have been a good fits for the larger screen. Maybe they just want to test the water and avoid putting all their eggs in the same basket.

I am really looking forward to experiment with those devices to see which kind of innovations and new User Interface paradigms you can create on those.

If you want to watch Microsoft short presentations on each device: Surface Neo (the larger one) and Surface Duo (the smaller one, based on Android).

HTTP/3 is coming faster than I expected.

If you would like to play with HTTP/3 today, Cloudflare is providing a nice way to do so: HTTP/3: the past, the present, and the future

I do not use Chrome or Cloudflare service, but they also provide an Open Source implementation for both server and client in Rust: Quiche: Savoury implementation of the QUIC transport protocol and HTTP/3.

Firefox is also going to provide soon a nightly build that support HTTP/3.

HTTP/3 is becoming a reality faster than I expected.

It looks like Dropbox will be more and more becoming a bloated all-in-one tool. I guess it is time to unsubscribe, as I do not like the way it is heading.