- Tram Ho
In recent years, with every new Windows update, users feel like they are gambling on the stability of their computer. Every new Windows update contains some nasty, sometimes serious, bugs. Many users also have to find ways to block automatic updates so they don’t damage the stability they have with their current version of Windows. Even in October 2018, Microsoft itself had to stop releasing Windows 1809 just five days after it launched because there were so many bugs.
What is the cause of this situation? No one can know exactly when Microsoft itself never mentioned this. But the answer by Jerry Berg, a former senior engineer in developing testing tools (Software Development Engineer In Test or Senior SDET) of Microsoft has partly revealed the reason for this situation.
Blue screen error haunts Windows.
Berg spent 15 years at Microsoft designing and developing tools and processes to automate Windows OS testing. He left the company after Windows 8.1 began shipping to users. After a few years, Microsoft began to significantly change the software testing process, and most likely this significantly contributes to the number of bugs occurring in every new Windows update.
Replace real people and computers with virtual environments
Before 2014/2015, Microsoft set aside a separate team for testing the operating system, builds, updates, drivers, and other code sets. This group consists of many departments that perform testing and during daily meetings they will discuss bugs and other arising issues. Tests are performed manually by this group in conjunction with automated testing tools, and if the tests pass, they will be accepted for the integration of lines of code into Windows.
Furthermore, the group’s tests were performed on “real” hardware in the lab under automated testing. These computers have different hardware components, for example different types of processors, types of hard drives, different types of video and sound cards, as well as a variety of system configurations. This means that errors that only occur with certain types of hardware or configurations appear during testing.
Testing Windows with real testers and computers has been replaced by Microsoft with virtual environments
However, when Microsoft changed direction with its focus on merging the three main components – Windows, Windows Mobile and Xbox – into a single system, the company laid off almost the entire Windows Test team. this. Now testing is mostly done on virtual machines, and according to Mr. Berg, this means the tests are not performed on real machines with the same variety of hardware configurations as before.
In addition, in the past, Microsoft employees were able to install Windows on their own computers. This also means that they can participate in the testing process as well. The main idea behind this is being able to get feedback from Microsoft employees as soon as they encounter a problem on the job. But now, Berg notes that the word install of Windows is no longer as widely used as it used to be.
Collect Windows error data using Telemetry
In addition to the automated test systems, other test data sources come from Telemetry and users of the Windows Insiders program. Windows Insider user builds are installed on millions of devices, and Microsoft collects Telemetry data from all of these devices.
Every time something crashes, Microsoft gets information about it. However, one of the problems with Telemetry is that it does not detect most other errors, thus rendering the collected data inaccurate. If something is not working properly, Microsoft also cannot distinguish it from other Telemetry data.
Windows updates are so buggy that many users choose not to update to avoid ruining the stability they currently have.
Theoretically, users could report these issues, but many people don’t, or report it at another time, maybe the issue has been drowned out by other responses from Insider users that Microsoft received. OK. Besides, while Insider users may report bugs, they often don’t say what is causing the problem. This caused major problems for the engineers tasked with solving them.
In the years 2014/2015, Microsoft’s testing team will be tasked with analyzing errors and problems that arise, and then providing engineers with the data they need to solve them. But according to Mr. Berg, today, engineers are left with only data from Telemetry to try to figure out how to fix those errors and problems. They will then be pushed back to the customer’s device running Insider builds to check if the bug has been fixed and or if new bugs have arisen.
Not just more errors per Windows update
One of the main reasons that Microsoft recently stopped releasing many new features at the same time to customers is because if problems are not detected in time, they will affect a large number of users. And the source of that is because the testing process has been shortened and the number of personnel involved is no longer as large as before.
If your update fails, update it again to fix the error.
Windows 10’s catastrophic update 1809 is a clear demonstration of how trying to roll out a bunch of new features without being able to test them correctly will have the consequences. Since that update, Microsoft has lost the courage to roll out many new features as before.
This is all because Microsoft has traded its test team with Telemetry data they gathered from Builds on Insider users, as well as replacing real computers with a variety of configurations. each other using virtual environments.
This is not only annoying and hesitant to update to a new Windows version as more and more errors appear, but also bored users when each update only offers a handful of new features. . In the end this makes the user even more reluctant to update – why fix something when it doesn’t break?
Refer to Ghacks
Source : Genk