End of Life for Azure Notebooks
github.com
github.com
The headline and this sentence seem contradictory
By 2015, we're talking about Windows Phone 10. After 7, 8, and 8.1, the world was pretty much done with Windows Phone. Microsoft pulled the plug less than two years later.
You could argue in a round about way they were killed because they didn't execute competently or something, but several things on that list were either outcompeted or obsoleted by their environment (Silverlight => HTML5).
-
The problem most people have with Google is when products that are doing well and are well loved by many... but not up to Google scale standards
Wunderlist fits that description. They did replace it with To Do, which was made by the Wunderlist team, but technically not the same app
XNA does, but MonoGame picked up the slack
And of course... Microsoft Reader. That hideous Metro app that would try and open in full screen back in the early Windows Store days. Good riddance.
Then the iPhone 3G was released, and IMO that’s when iOS showed how wart-filled WinMo had become.
https://www.versionmuseum.com/history-of/discontinued-micros...
Microsoft Teams Progressive Web App https://www.versionmuseum.com/history-of/discontinued-micros...
And also, it list things that were consolidated into other apps, I don't know you could say it was killed off, but more moved. (such as Microsoft groups, is just a part of outlook now, and not a standalone app)
I get the sense they're taking a "Manhattan project" approach to grabbing cloud share. They find every promising technology they can, and either try to re-implement in their ecosystem or support it.
Some of the stuff they're throwing at the wall sticks, and some of it doesn't.
Since the introduction of .NET, when you seen Microsoft going all .NET, or having a cycle talking about why native is good and .NET should only be for LOB, it reveals which division is winning the political shots.
Some examples, Longhorn's failure, only to redo many of its ideas with COM in Vista, which ended up in WinRT.
WinRT/UWP looks suspiciously similar to ideas that gave birth to .NET, before it was decided to create a managed runtime instead (Ext-VOS) project.
Midori also proved it's capabilities by running part of Bing for a while, but it wasn't enough.
Joe Duffy has a remark on its Rustconf keynote, that even when proven wrong and given the opportunity to test drive Midori, the Windows team wasn't willing to change their mind.
Then you have them killing C++/CX in name of ISO compatibility, although UWP is anyway Windows only. So now DevDiv has to wait for the day ISO C++ supports the reflection and metaclasses, so that they can replicate the C++/CX tooling experience introduced in 2012.
Apparently they already shut down 2 weeks ago.
Any word on what's replacing it? Azure Machine Learning notebooks?
I didn't know about GitHub code spaces with Jupyter support.
I really wish they would have kept it alive much longer... Azure Machine Learning is way more complex and Codespaces not yet available for normal people.
Only free alter alternative with private Notebooks is Colab.
Paperspace Gradient is awesome, but the free version requires public Notebooks.
- No setup Jupyter notebooks
- Collaborative notebook editing
- Publish notebook into AppBooks (parametrized notebooks) that allow you to deploy the resulting model, track experiments, etc. This also enables people unfamiliar with Notebooks to use it as an application without being overwhelmed, and tweak models.
- Schedule notebooks and see their output live, even if you close the browser. Your notebooks are run on a much beefier machine than when you're in interactive mode. You won't have the problem of a training job where you don't get the result because of a lost connection.
- Automatic model detection and saving, parameter/metrics tracking
- One click model deployment
- Multiple notebook checkpoints (revert back and forth to previous saved versions)
Caveat emptor: as you may notice, we're not focusing on embellishing notebooks as we believe, from what we have experienced, that the fact machine learning projects take a long time is not for lack of beautiful stylesheets. We'll tackle these problems at a later time.
- [0]: https://iko.ai
- [1]: https://www.reddit.com/r/learnmachinelearning/comments/je0pm...
The guys I talked to during the process were all pretty cool, hope they land on their feet.