Finishing | Zheng Liyuan
Produced by | CSDN（ID：CSDNnews）
I believe that programmers engaged in Internet product development are no strangers to the launch of projects, after all, products need to be constantly iteratively updated. Therefore, the launch is more grand: R&D personnel must be present, the test team is indispensable, the operation and maintenance team is even more important, and product managers, designers, etc. also need to accept the online content......
Considering the complexity of the entire operation process and follow-up work, each company must have corresponding systems and specifications for product launch, and generally speaking, there is also a fixed launch date, which may be every Wednesday or Friday - so ask: As a product development programmer, what day of the week do you participate in the project launch?
In response to this problem, the chief engineer of Octomind, an AI testing tool, recently issued an initiative: "Never release an update on Friday!" and this point of view quickly attracted great attention and discussion.
The popular Octomind has been updated 64 times, 9 of which were on Friday
Before we get into the nitty-gritty, let's take a quick look at Octomind. It's an end-to-end, browser-based testing tool that automates maintenance and builds, integrating with Github Actions, Azure DevOps, and more. As you can see from Octomind's official website, the main focus is to free up the testing time of engineers: "What is the first thing that engineers would most like to spend less time doing? Don't waste time on the wrong tests, just develop the build. ”
According to the website, users only need to provide Octomind with a URL, and it can use AI to discover, execute, and maintain its E2E (end-to-end) tests. The so-called E2E test is to simulate the operation process of real users, starting from the user interface, through various levels of system components to the final output, so as to verify whether the function and performance of the entire system in all aspects are qualified, and ensure the quality and stability of the product.
However, as Octomind points out, there is a major trust problem with E2E testing, which is that code bugs are not the only cause of test failure: third-party dependencies, timing issues, randomness, race conditions, etc., can lead to unstable and unreliable test results.
The result is Octomind, which is designed to help developers save valuable time "debugging perfectly healthy code" and is plug-and-play for CI/CD pipelines, providing timely feedback of test results to users and detailed information for users to review, review, and debug.
With these features, Octomind has received a lot of praise from industry insiders:
- Duc Tam Nguyen, co-founder and CTO of Aimino, said, "With Octomind, it's like two more people on my team delivering high-quality test cases, and it works better than any other tool we've tried";
- Aditya Advani, Co-Founder and CTO of Best Parents, also praised Octomind: "As an early-stage consumer market, we need to maintain high quality and speed. Luckily, Octomind came in. At this moment, a happy engineering leader gave a thumbs up. ”
According to statistics, the popular Octomind has been updated 64 times through November of this year, and 9 of them have been on Fridays - however, recently Octomind Principal Engineer Daniel Draper wrote a blog post titled "Never release an update on a Friday".
Refusing to go live on Friday means the code isn't good enough?
Daniel Draper blogged about the common belief among many programmers in the software development industry that Fridays are not appropriate to release software: Friday's deadlines often push them to rush things through and may force releases without adequate checking, leading to problems.
- Impaired quality: Tendency to omit seemingly unimportant tests or ignore the results of failed tests, resulting in an unstable release.
- Weekend overtime: If something goes wrong, developers have to give up their weekend time to fix the problem, leading to burnout and dissatisfaction.
- Forgetting details: As a developer's focus shifts, they may not be able to recall all the details of the go-live, making troubleshooting more difficult.
However, Allen Holub, a well-known computer writer, takes the opposite view: "If you release a few times a day, with minimal delays, no known bugs allowed at the time of release, a lot of testing (mostly automated), and multiple people reviewing the code as they write, then the risk of releasing on a Friday is essentially zero." ”
Allen Holub has even said bluntly that if you refuse to release your code on Friday, then your code is not good enough - Daniel Draper has a different opinion: "I don't agree with Allen's statement that 'not releasing on Friday' means not being of high quality. ”
In fact, at the beginning of the blog, Daniel Draper explained that the team itself does not refuse to release updates on Fridays, but rather values the developer's time off.
As mentioned above, the team has released product updates on Fridays 9 times, but bugs keep popping up unexpectedly, regardless of whether it's a weekday or not. And his suggestion of "don't go live on Fridays" is just a matter of cherishing the weekend for himself and other engineers: "Fundamentally, it's just a matter of setting expectations for the team. Then if there is an opportunity to reduce the possibility of rolling back overtime on weekends or fixing bugs, no matter how small the opportunity, I will take it. ”
Daniel Draper also added that bugs should be found and fixed before a product goes live, but no one can 100% guarantee that there will be no bugs, so "if there's no business to have to be released on Friday, I don't think it's a problem to push a bigger version to Monday."
Attracting hot attention, Octomind collapsed for a while
Daniel Draper's blog has caught the attention of many developers on Reddit and has resonated with many:
- "That's right, bugs can happen at any time, but if they happen on Fridays, it's even worse for the team. That's the way I'm going, I generally avoid going live before work because it can be a headache to deal with if something goes wrong. ”
- "That's why our team doesn't typically release products after 3 p.m. on Fridays or weekdays. Because our experience is that there are usually only two problems that occur after a release: it crashes immediately and then rolls back, or it works fine for a few hours or even a day, but then you find that the performance is degraded or some of the data is corrupted. Therefore, it's best not to go live before hours or on Fridays. ”
- "So, there's a reason why the entire gaming industry releases/updates on Tuesday. On Monday, make sure that everything you produced last week is organized and ready to be pushed. It is then released on Tuesday, and bug fixes can be made until Friday, with little to no need to worry about it on the weekend. ”
However, some have pointed out that certain products are more appropriate to go live on Friday, such as SaaS and games:
- "A lot of SaaS vendors that focus on larger businesses release software on Fridays because they cater to the 9-to-5 crowd that doesn't typically work on weekends. I'd rather deal with a bug that affects a transaction when the volume is low than when the volume is high. ”
- "I'm serious, the only software industry that should consider releasing any updates on Fridays is the gaming industry, as their customers have weekends off. Generally speaking, they are happiest when they can play new games all weekend. ”
Another developer suggested that the best time to launch a product is on Wednesday: "Generally speaking, everyone is in the office on Monday, but they often haven't come back from the weekend, and this state of affairs may continue until Tuesday. So in my opinion, Wednesday is the best day to launch the product, when the engineers are already in the groove and there is Thursday and Friday to fix bugs. ”
Eventually, the unexpected popularity of Daniel Draper's blog got many people interested in Octomind, an AI testing tool, and tried to sign up for it—only to have Octomind crash due to the sheer number of people signing up, putting too much load on its back-end servers running in the cloud.
Of course, Octomind has fixed this issue so far, and they have learned a lesson from it: "We try not to publish it on Friday, and maybe we shouldn't publish that blog either." ”
So what day of the week does your team go live, and what do you think about when the product goes live?
Reference Links:
https://www.octomind.dev/blog/we-went-viral-with-a-broken-app
https://www.reddit.com/r/programming/comments/18dprj9/never_ship_on_fridays/