Mi Pad 5 details exposure: locked for August release; WebStorm 2021.2 officially released | SiFu Weekly

SegmentFault Thought

40s News Express

• Tencent Sogou completed preliminary integration in August, a large number of employees resigned: dissatisfied with demotion and salary adjustment
• Didi responds to privatization: false news, is cooperating with cyber security review
• 24,999 yuan, Huawei V75 Super smart screen released: adopt self-developed Honghu SuperMiniLED solution
• Apple's new technology exposure: iPhone 14 is expected to support under-screen fingerprints and under-screen Face ID
• OPPO and vivo will soon launch self-developed chips, and all first-tier mobile phone brands will enter the market to make cores
• Mi Pad 5 details exposure: locked for August release
• Facebook shares its experience of migrating to MySQL 8.0
• Raspberry Pi launches recycling program, which will be extended to the world in the future
• Malware prefers to use "unpopular" languages ​​such as Nim, DLang, and Rust
• WebStorm 2021.2 is officially released, the UI interface is completely Chinese, and many new features have been added
• Sony PS5 open built-in SSD interface: support up to 4TB M.2 hard drive
• Two years later: Google UI tool Jetpack Compose finally ushered in the 1.0 official version
• OpenAI open source new AI programming language to replace Nvidia CUDA
• Atom 1.58.0 released, the official GitHub text editor

Industry Information

Tencent Sogou completed preliminary integration in August, a large number of employees resigned: dissatisfied with demotion and salary adjustment

Tencent will initially complete Sogou integration work in August at the earliest. Most of Sogou's business will be merged into Tencent PCG, and the main team will be merged with Tencent Kandian; under the structure of Tencent Kandian, several new centers will be established for some of Sogou's teams. Sogou executives will assume the positions of general manager (GM) and director. In the future, after Sogou completes the overall delisting, the current CEO Wang Xiaochuan may leave Sogou and choose to start his own business; and Yin Yu, vice president of Tencent Group and head of Tencent Kandian, will become the new leader of the Sogou team.

It should be noted that in many people’s eyes, Sogou’s merger with Tencent is a good thing for employees. It is possible to join Tencent in this way and enter one of China’s largest Internet companies, but reports point out that there are currently about 4 or 5 A hundred people chose to leave Sogou, accounting for nearly a quarter.

According to the news, some employees left because they were dissatisfied with the changes in their ranks after joining Tencent, because after being merged into Tencent, Sogou's original human compensation system will also be fully adjusted against Tencent.

Among them, the ranks of some Sogou employees will be lowered by 2 to correspond to the rank system after entering Tencent. For example, the vice president becomes GM and the director becomes group leader. This is the most unacceptable point for many old employees, so they chose to leave Sogou. .

However, it should be noted that the news said that although the ranks of these people have been lowered, their salaries have been generally raised. The employees of the original Sogou team whose salary is between 14.6 and 14.7 months are uniformly raised to 16 salary per year. For some employees, it was a surprise.

Didi responds to privatization: false news, is cooperating with cyber security review

It was previously reported that after the escalation of regulatory actions, Didi began to consider a privatization plan around mid-July. Didi is also considering introducing government-backed investors who may fund the transaction and help guide the company as Didi tries to solve data security issues. Didi denied the news, "The rumors about Didi's privatization are false information, and the company is currently actively cooperating with cybersecurity reviews."

24,999 yuan, Huawei V75 Super smart screen released: adopt self-developed Honghu SuperMiniLED solution

News on July 29. At tonight’s press conference, Huawei released a number of new products, such as Huawei’s P50 series mobile phones, Huawei’s smart screen V98 giant screen, etc. Among them, Huawei’s smart screen V75 Super is Huawei’s first MiniLED smart screen with a price tag. 24,999 yuan, officially announced at 00:00 on August 12th.

Huawei Smart Screen V75 Super adopts Huawei's self-developed Honghu Super MiniLED precision matrix backlight solution. There are 46080 MiniLED beads inside the screen, each of which is only 1/120 of the area of ​​traditional LEDs. It also has 2880 physical backlight partitions and supports 3000nits. HDR peak brightness, more true to restore natural colors.

In addition, Huawei’s smart screen V75 Super is equipped with a 20-unit Deware audio system with 75W output power. The speaker units are distributed around the screen to bring a cinema-level 3D surround sound field.

In other respects, the machine supports the HDR Vivid standard, adopts QD quantum dot technology, and covers 97% of the DCI-P3 color gamut. This smart screen is equipped with Huawei Hongmeng HarmonyOS OS 2 system.

Apple's new technology exposure: iPhone 14 is expected to support under-screen fingerprints and under-screen Face ID

According to the current known news, Apple will officially release the new iPhone 13 series in September this year, but unfortunately this machine did not remove the bangs that have plagued users for many years. Instead, it optimizes the internal stacking to remove the bangs. The area of ​​the area has shrunk.

It is reported that the main reason why Apple has not started with Liu Hai for many years is that it has not yet found an effective alternative to Face ID and a sufficiently secure unlocking method. At present, the 3D structured light of Face ID is still one of the most secure unlocking methods. one.

But this situation may soon be broken. According to related reports, Apple has a new technology patent exposed, which is exactly a method of placing a sensor under the screen.

The patent points out that the iPhone can receive the light transmitted through the screen in a direction by placing the optical imaging array below the screen (the opposite of the direction in which the screen emits light), and this optical imaging array can be applied to any imaging , Sensing, or data aggregation, including but not limited to biological characteristics such as ambient light, structured light, fingerprints, and facial features.

In other words, the iPhone equipped with this technology is expected to realize fingerprint recognition under the screen, Face ID face recognition under the screen, and even iris recognition under the screen. This technology will also enable the iPhone It is hoped that the bangs area can be completely erased, which can achieve a true full-screen effect and bring a more shocking visual experience.

In view of the current iPhone 13 series has entered the mass production stage, its design has been finalized and will not be equipped with this technology, and next year’s iPhone 14 should be about to enter the research and development cycle and is expected to become the first model equipped with this technology.

Therefore, iPhone 14 will also hopefully become the first Apple mobile phone in history that supports under-screen fingerprints and under-screen Face ID, which is worth looking forward to.

OPPO and vivo will soon launch self-developed chips, and all first-tier mobile phone brands will enter the market to make cores

According to several people familiar with the matter, after Xiaomi launched its self-developed surging C1ISP (Image Signal Processor) chip, two other domestic first-tier mobile phone manufacturers, OPPO and vivo, are also about to release self-developed ISP chips. An OPPO insider said that the OPPO self-developed chip project has been advancing, and the team currently has about a thousand people. The first product is an ISP chip similar to Xiaomi's surging C1, which will be launched on the Find X4 series of mobile phones that will be launched early next year. .

Mi Pad 5 details exposure: locked for August release

Recently, the well-known overseas media 91mobiles brought the detailed specifications of the Mi Pad 5, which showed that there are three models of Mi Pad 5: high, medium, and low. The high version will be equipped with the Snapdragon 870 chip. Equipped with Snapdragon 860 and Snapdragon 768G processors.

It’s worth mentioning that it was previously reported that, except for the core configuration of the Mi Pad 5, the engineering machine has the same other parameters, which means that these three models will use the same size screen. The main difference is in performance. aspect.

The news revealed that Xiaomi Mi Pad 5 has an ultra-high-quality LCD display with a size of 10.95 inches (called 11 inches), a resolution of 2560*1600, and a 120Hz refresh rate, which can provide an extremely delicate and smooth display effect. , Which can effectively improve the user experience for large-screen devices.

At the same time, this news also revealed that the Mi Pad 5 will have a built-in 8720mAh large battery, and support 67W wired fast charging, which can quickly replenish power, which is a lot ahead of the iPad.

Facebook shares its experience of migrating to MySQL 8.0

In a blog, Facebook shared the company's experience in a somewhat difficult large-scale leapfrog migration to MySQL 8.0 version. Previously, it has been using MySQL version 5.6.

MySQL is an open source database developed by Oracle that supports some of Facebook's most important workloads. According to Facebook, each new major version of MySQL requires a lot of time and effort to migrate workloads. Among the challenges are:

• Port its custom functions to the new version
• Ensure replication is compatible between major versions
• Minimize the changes required to query existing applications
• Fix a performance regression that prevented the server from supporting its workload

According to reports, it took more than a year for Facebook to upgrade to MySQL 5.6 last time; and this upgrade to MySQL 8.0 also took several years. When version 5.7 was released, Facebook was still developing the LSM-Tree storage engine MyRocks on version 5.6. Since upgrading to 5.7 while building a new storage engine will significantly slow the progress of MyRocks, the team chose to continue using 5.6 until MyRocks is completed. MySQL 8.0 was released just when MyRocks was completed, so Facebook chose to upgrade to improve its storage engine.

Facebook pointed out that migrating to 8.0 is significantly more difficult than migrating to 5.6. They have 1,700 code patches to migrate from their customized MySQL 5.6 branch to 8.0. Due to the new features of Facebook's MySQL and the fixes that are constantly added to the 5.6 code base, this work has become very complicated.

Because the upgrade from 5.6 to 8.0 completely skipped 5.7, some APIs that were active in 5.6 were either deprecated or completely deleted; this means that any application that uses the old API needs to be updated. And some features of Facebook are not forward compatible with similar features in 8.0, and need to be deprecated and migrated forward.

There is also the problem of uneven custom code documentation. Facebook claims that most of its custom code is well-commented and documented. But other codes are not well documented. Facebook needs to dig out old files, posts, and code comments to understand the history.

In the end, Facebook evaluated more than 2,300 patches and ported 1,500 of them to MySQL 8.0. "We have converted many InnoDB replica sets to run completely on 8.0. Most of the rest are in different stages of the migration path. Now most of our custom features have been migrated to 8.0, and the update to Oracle's minor version is relatively Easy, we plan to keep up with the latest version."

"Despite the obstacles we have encountered on the road to migration, we have seen the benefits of running 8.0. In general, the new version greatly expands what we can do on MySQL @ Facebook."

Raspberry Pi launches recycling program, which will be extended to the world in the future

It has been 9 years since the Raspberry Pi was launched in 2012. Coupled with the low price of Raspberry Pi products, many developers and geek users will buy and upgrade with the release of new products. . When new products are used, the old products that have been retired can only "eat ashes" in a certain corner of the house.

OKdo has proposed a solution to this wasteful situation. It has cooperated with Raspberry Pi and Sony Technology Center to jointly create the OKdo Renew plan, which is an official plan to update the old Raspberry Pi to the new one. It provides Created an economic incentive to make people willing to use this service.

OKdo Renew plans to recycle Raspberry Pi3b, Pi3b+ and Pi4 in the hands of users. In return, users will get a voucher worth 10 pounds, which can be used to deduct the next time they buy a new product at OKdo. However, this voucher also has conditions for use. First, it can be used when the order amount exceeds 15 pounds, and secondly, the voucher is valid for 30 days.

Eben Upton, the founder and CEO of Raspberry Pi, said: "The Raspberry Pi has a long-term commitment to sustainable development. The Raspberry Pi consumes less electricity than traditional PCs, and consumes more energy, materials, and labor in the manufacturing process. Less. Since 2012, more than 40 million Raspberry Pi units have been sold, most of which are still in working condition. Refurbishing and reusing old equipment can further reduce the carbon footprint."

Initially, this service will be promoted in the UK, the birthplace of the Raspberry Pi, and then plans to expand the service globally by the end of 2021.

The recycled Raspberry Pi will be sent directly to the Sony Technology Center where they were originally manufactured, where the Raspberry Pi will be fully tested, repaired and updated, using recyclable packaging, and finally resold through OKdo at a lower price. OKdo will provide a 12-month warranty for the refurbished Raspberry Pi, and will donate a portion of each sale to the Raspberry Pi Foundation. Those Raspberry Pis that cannot be refurbished will be processed in the Sony Technology Center to reduce the resulting waste.

Malware prefers to use "unpopular" languages ​​such as Nim, DLang, and Rust

Security researchers have found that malware authors are increasingly using relatively unpopular programming languages, such as Go, Rust, Nim, and DLang, to create new tools and hinder researchers from analyzing.

According to a report released by the BlackBerry Research and Intelligence team on Monday, the use of the above-mentioned four languages ​​in the malware family has seen a significant increase. The team chose these four languages ​​for research, partly because they fit their detection methods, and because these languages ​​have strong community support and have better development prospects. The report pointed out that these relatively unpopular programming languages ​​are no longer as rarely used as once thought, and malware authors have begun to use them to rewrite more well-known malware or create tools for new malware.

The reason behind this is that malware authors hope to use unpopular programming languages ​​to circumvent the analysis and detection of the security community and solve some of the pain points in the development process.

Specifically, security researchers tracked more loaders and releasers written in popular programming languages. The malware Remcos, NanoCore Remote Access Trojans (RATs) and Cobalt Strike mentioned in the report use this approach, so it is difficult for mainstream security analysis methods to discover preliminary and advanced malware deployments. Researchers say these methods are often used to help malware evade endpoint detection.

This report also describes the trends in the use of these languages ​​in malware families. In the past few years, Go is the most used language, and many malwares are written based on it. But in the near future, the adoption rate of Dlang and Nim is also gradually increasing. In this regard, the researchers believe that by using less common programming languages, tools created by malware developers will be more difficult to reverse engineer by security analysts, thus hindering their research.

The researchers also mentioned that malware authors are avoiding the use of signature-based detection tools to improve cross-compatibility of target systems.

The BlackBerry research team warned that it will take some time for malware sample analysis tools to catch up with these "new" languages, but the security community "must proactively defend against the malicious use of emerging technologies." Eric Milam, vice president of BlackBerry threat research, also said that the industry and customers must understand and pay close attention to these trends, because malware will only continue to increase.

The latest technology trends

WebStorm 2021.2 is officially released, the UI interface is completely Chinese, and many new features have been added

WebStorm 2021.2 is officially released, and the updates are as follows:

code editing

Reload page when saving

WebStorm can now automatically update the page in the browser when editing and saving your HTML, CSS, and JavaScript files. To get started, open an HTML file in the editor, hover your mouse over it, and click the icon of the browser you want to use-all browsers support it.

Operation when saving

Do you like to perform certain operations when saving? WebStorm 2021.2 now supports this workflow correctly. We have redesigned all existing features, collected them in one place, and enhanced it with some new options, including the ability to reformat the code when saving and optimize the import .

Create scratch files faster

Want to deal with some code outside the context of the project? Select it in the editor, then press ⌥⏎, and choose to create a new scratch file from the selection. This will create a scratch file with the required code.

Code With Me

When you are in follow mode in a Code With Me session, you can now track the code completion suggestions used by the people you follow.

JavaScript & TypeScript

Automatic import require()

WebStorm can add missing import statements when you complete ES6 symbols, and it can now do the same for CommonJS modules-require imports can be inserted during code completion.

Support TypeScript type

WebStorm now correctly supports TypeScript syntax used in .js files. We redesigned and expanded existing support and fixed many known issues.

New operation of arrow function

Need to quickly add an arrow function to your code? Now you can use ⌥⌘J to surround a code block with an arrow function. You can also enter arf and press ⇥ to expand the template and add an empty arrow function.

Preview option page

The preview options page used to work only in the project view, but now it also works when you debug your application. You can turn on this feature in the Preferences/Settings|Editor|General|Editor tab. This will help you avoid messing up the editor with multiple files, because WebStorm will open these files consecutively in one tab.

Frame and technology

React useState hooks

You no longer need to waste time refactoring useState values ​​and functions one by one-WebStorm can now rename both for you. Place the cursor on a status value, and then press ⇧F6 or enter Refactor | Rename from the right-click menu.

for class names and clsx libraries

To help you use CSS classes more effectively in React projects, we have added support for popular class names and clsx libraries. WebStorm will display your CSS class completion suggestions and resolve all symbols with literal names in string literals and attributes.

General support for network types

We have extended support for web-types, which is an open source standard for documenting web frameworks. It used to mainly support Vue. But now you can use it to enrich the coding assistance of custom components in your HTML files.

Version control

New pre-commit check

WebStorm has added a new option that allows users to run tests to check the code before submitting. Click the gear icon in the submission tool window, select Run Test, and select the desired run configuration. WebStorm will test your files and warn about any problems.

GPG signature support

Now you can protect your submission by signing it with a GPG key. You can turn on this feature by configuring the GPG key in Preferences/Settings|Version Control|Git.

Local history search

Viewing revisions in the local history is now easier. Right-click on the modified file, then enter the local history|display history, and use the search bar to browse your modifications.

Availability

Localized user interface

Starting from this version, you can enjoy the fully localized WebStorm user interface in Chinese, Korean and Japanese. Localization can be used as a non-bundled language pack plug-in, which can be installed from Preferences/Settings|Plugins in WebStorm.

Faster tool window rearrangement

It is now easier to rearrange the tool windows. Hover the mouse over the top of the tool window you want to move, and then drag it to the place you want. If you want to separate a tool window from the main WebStorm window, you can do the same-just drag it out of the IDE frame.

Customize project icons more easily

WebStorm makes it easier for users to assign custom icons to projects. Right-click a project on the welcome screen of WebStorm, then select the option to change the project icon, and upload the SVG file you want to associate with the project.

from the toolbox application

You will no longer miss any key product updates for the Toolbox app. If a new version is available for download, WebStorm will notify you and let you choose to upgrade to that version-just make sure you have Toolbox App version 1.20.8804 or higher.

Improved preferences/settings dialog

In your preferences/settings dialog, there is now a node for advanced settings. You can find some new configuration options there, including the ability to set the left margin in interference-free mode. In addition, you can now jump faster between recently opened nodes-just use the arrow in the upper right corner of the dialog.

Automatic cache and log cleaning

After each major update, WebStorm will clean up any cache and log directories older than 180 days from the last update. The system settings and plug-in directory will remain the same. To manually trigger this process, please go to Help|Delete the legacy IDE directory in the main menu to set it up.

New functions of the built-in terminal

Three new functions have been added to the built-in terminal to improve the user experience. You can now change the shape of the cursor and use the ⌥ key as a meta modifier-look for these options in Preferences/Settings|Tools|Terminal. In addition, it is now more convenient to use http links.

Sony PS5 open built-in SSD interface: support up to 4TB M.2 hard drive

Sony has pushed the latest Beta version of the firmware for PS5, unlocking the expansion capability of the built-in idle M.2 interface. Users can disassemble the shell of the PS5 by themselves and install the M.2 NVMe solid state drive independently.

It is understood that the PS5 is compatible with PCIe Gen4 x4 M.2 NVMe SSDs with a capacity compatible with 250GB-4TB. The specifications of the solid state drive are compatible with M.2 2230, 2242, 2260, 2280, and 22110.

It should be noted that PS5 is only compatible with products with a width of 22mm PCB. Enterprise-level hard drives with a width of 25mm are not supported. However, the width of the heat sink can reach 25mm, and the total thickness of the heat sink should be less than 11.25mm in terms of height. The thickness from the front surface of the PCB to the highest point must be less than 8mm.

In addition, Sony also supports the transfer of games to a newly installed SSD. Users can download games from the game console store to the M.2 SSD, go to the main screen, and select "Settings">Storage>Installation Location to set this setting.

Specific steps are as follows:

1. Go to your game house and select the game library.

2. Highlight the data to be moved, press the "Options" button, and then select "Mobile Game" and "Apply".

When you use both M.2 SSD and USB expansion storage drive at the same time, go to the item that can be moved to the M.2 SSD storage tab.

3. Select the checkboxes and select "Mobile" for any other games and applications you want to move.

To check your storage space, and move or delete data on the M.2 SSD, go to the home screen and select M.2 SSD storage>Storage>Settings.

Two years later: Google UI tool Jetpack Compose finally ushered in the official version 1.0

Since its launch at the I/O 2019 Developers Conference, Jetpack Compose is destined to become the officially recommended Android UI development method. After two years of continuous development, we have finally reached a milestone-on July 29, Google officially launched version 1.0 of Jetpack Compose.

Android Police pointed out: In addition to the stable version of Android Studio Arctic Fox, Jetpack Compose can also be used in a production environment.

In March of this year, Google released the 1.0 test version of Jetpack Compose, and actively identified and fixed related bugs in the following months.

At the same time, we have also ushered in some newly added features, such as Compose Preview and Deploy Preview.

● The former allows developers to view the real-time rendering of Compose UI in different states, such as light and dark mode or different font scaling.

● The latter pays more attention to the functional level, allowing developers to push updated code to users without completely restarting the application.

Both of the above require Android Studio Arctic Fox, which also ushered in a stable version update today.

Before integrating Jetpack Compose into a newly developed App, many people are looking forward to the 1.0 stable version. The framework allows Compose to be used in existing views or included in Composable.

This allows developers to integrate Jetpack Compose into their projects at any speed without the need for large-scale conversions or even any code rewriting.

In addition, Compose can be smoothly integrated with other Jetpack libraries to implement functions such as LiveData and Navigation.

All UI elements are designed using Material Design and support theming, so developers can customize the appearance of the App and use rich animation APIs to convey various activities.

OpenAI open source new AI programming language to replace Nvidia CUDA

OpenAI announced the launch of Triton, an open source programming language similar to Python, and released version 1.0 of Triton. According to the introduction, Triton enables researchers without CUDA experience to write efficient GPU code, and in most cases it can be comparable to the quality of code written by experts.

OpenAI claims that Triton makes it possible to reach peak hardware performance with relatively little effort. For example, it can write an FP16 matrix multiplication kernel that matches the performance of cuBLAS in less than 25 lines of code, which is not possible for many GPU programmers. "Our researchers have used it to generate cores that are 2 times more efficient than equivalent Torch implementations. We are happy to work with the community to make GPU programming easier for everyone."

OpenAI pointed out that some of the problems considered to be quite challenging when using the CUDA framework have been solved in Triton. Compared with coding in CUDA, Triton can provide significant ease-of-use advantages for some neural network tasks at the core of the AI ​​machine learning form (such as matrix multiplication).

The Triton language automatically performs many AI code optimizations to save developers time; its relative simplicity also enables software teams without extensive CUDA programming experience to create more efficient algorithms than other methods.

OpenAI scientist and project leader Philippe Tillet told ZDNet that their goal is to make Triton a viable alternative to CUDA in deep learning. "Triton is for machine learning researchers and engineers who have good software engineering skills but are not familiar with GPU programming." Triton first appeared in the public eye in a paper published by Tillet in 2019.

According to the introduction, Triton improves AI performance by optimizing the three core steps of the data processing workflow of the machine learning algorithm running on the Nvidia chip:

The first step is the task of moving data between the GPU's DRAM and SRAM memory circuits. The GPU stores information in DRAM when it is not active and transfers it to SRAM memory to perform calculations. The faster the data transfer between the two components, the faster the machine learning algorithm will run, which is why developers prioritize optimizing this aspect of the computing workflow as part of the AI ​​project. The optimization process involves merging data blocks moved from DRAM to SRAM into large information units.

OpenAI said that Triton will automate tasks to save developers time.

The second calculation step of Triton's optimization is to allocate the incoming data blocks to the GPU's SRAM circuit in order to analyze them as quickly as possible. One of the main challenges involved in this step is to avoid so-called memory bank conflicts, which are situations where two software accidentally try to write data to the same memory segment. Memory bank conflicts hinder calculations until they are resolved; this means that by reducing the frequency of such errors, developers can speed up the performance of their AI algorithms.

Tillet explained: “Data must be manually stored in SRAM and managed before reuse to minimize shared memory bank conflicts during retrieval.”

The third and final task, Triton helps to automate; what is involved is not the storage unit of the GPU, but its CUDA core, which is the calculation circuit responsible for calculating the data stored in the memory. There are thousands of such circuits in a single Nvidia data center GPU. They allow the chip to perform a large number of calculations simultaneously.

In order to maximize the performance of the AI ​​model, developers must configure it to distribute calculations across multiple CUDA cores; so that they can be done simultaneously, rather than one after another. Triton can also automate this task, albeit only partially. The reason it does not automate the entire workflow is because OpenAI tries to give developers the flexibility to manually customize the process for their projects as needed.

Tillet revealed that the team has used Triton to accelerate and rewrite most of its GPU cores, and is committed to making it more widely applicable through subsequent versions. It is worth noting that the software is currently only applicable to Nvidia GPUs; it cannot be used on AMD GPUs, nor can it be compiled to CPUs.

Atom 1.58.0 released, the official GitHub text editor

Atom is an open source text editor developed by GitHub. It supports macOS, Windows and Linux operating systems, supports plug-ins written by Node.js, and has a built-in Git version control system provided by Github. Most extension packages are open source licenses and are built and maintained by the community.

Atom is based on Electron (originally called Atom Shell), a cross-platform application framework that uses Chromium and Node.js, and is written in CoffeeScript and Less. Atom can also be used as an IDE. It is called "hackable text editor for the 21st Century" by developers. Since May 6, 2014, Atom's core programs, package manager, and Atom's Chromium-based desktop program framework are all released under the MIT license agreement.

Atom 1.58.0 is officially released. The notable changes in this update are as follows:

• 22315-Update to macOS Big Sur style icon.
• 22424-Fix the problem of reopening the project in safe mode and development mode.
• 22123-Improve the contrast of Windows application icons.
• atom/archive-view#73-Added the function of folding compressed archive directories (zip, tar, etc.).
• atom/bracket-matcher#405-Handle the multi-cursor selection problem in brackets.
• atom/find-and-replace#932-Add the right-click menu options of "Open in New Tab" and "Open in New Window"