They are:doers: self-motivated or intrinsically driven to achieve and learndrivers: proactive and decisive; don't wait for orderspromoters: passionate and excited to share what they create for their own benefit and the benefit of others

Review: a formal assessment or examination of something with the possibility or intention of instituting change if necessary.

On top of that, there’s one thing you can do to extend your life. By studying the philosophies of those who came before you, you absorb their experiences. Every philosophy book you read, you’re adding the author’s lifespan to yours. There’s no better way to spend your time than studying philosophy.13

What I was trying to say before was just because someone’s always busy, and lives to an old age, doesn’t mean they’ve lived long. They’ve just existed long.

I’m not saying you should lay down on the beach all day. I’m saying you should find something that’s enjoyable to you, and valuable for the world.10 You should live your life intentionally, instead of having your time stolen from you little by little.

Even with all that, he was looking forward to the day that he could step down and retire from it all. The man with all the power in the world was happiest when he thought about the day he could let go of all the power.8 How foolish is it to spend your life chasing fame, riches, and power, while being unhappy the entire time, even after you achieve it?

The most surprising thing is that you wouldn’t let anyone steal your property, but you consistently let people steal your time, which is infinitely more valuable.2

sudo killall coreaudiod

This is a neat Docker trick for those who have an ARM development machine (Apple M1), but sometimes need to build x86/amd64 images locally to push up to a registry.

So, we can create this builder on our local machine. The nice part about this creation is that it is idempotent, so you can run this command many times without changing the result. All we have to do is to create a builder profile and in this case I have named it amd64_builder.

sum(1 to n) = n(n + 1) / 2

Python 3.11 is up to 10-60% faster than Python 3.10. On average, we measured a 1.25x speedup on the standard benchmark suite. See Faster CPython for details.

But to my knowledge, this is the first time anyone has collected public info about Bigco dev environments in one place

Git also has a built-in command (maintenance) to optimize a repository’s data, speeding up commands and reducing disk space. This isn’t enabled by default, so we register it with a schedule for daily and hourly routines.

On a new clone of the Canva monorepo, git status takes 10 seconds on average while git fetch can take anywhere from 15 seconds to minutes due to the number of changes merged by engineers.

Over the last 10 years, the code base has grown from a few thousand lines to just under 60 million lines of code in 2022. Every week, hundreds of engineers work across half a million files generating close to a million lines of change (including generated files), tens of thousands of commits, and merging thousands of pull requests.

The perpetrator in question was completing an internship and committed code into the Windows 3.1 code base that was a little prank for the test team: Under a very specific error condition, it changed the index finger pointer to a middle finger.

23.0G com.txt # 23 gigs uncompressed

What makes you unique is not your specific attributes. It’s your specific ties to the network around you.

Becoming disentangled from your web of mutual commitments, shared history, and collective responsibility is to be rendered into a transaction, a slave.

Real friendships don’t form via shared interests. They form via shared context.

In a Staging workflow, releases are slower because of more steps, and bigger because of batching.

For Staging to be useful, it has to catch a special kind of issues that 1) would happen in production, but 2) wouldn’t happen on a developer's laptop.What are these? They might be problems with data migrations, database load and queries, and other infra-related problems.

Another disadvantage of managed platforms is that they are inflexible and slow to change. They might provide 80% of the functionality we require, but it is often the case that the missing 20% provides functionality that is mission critical for machine learning projects. The closed design and architecture of managed platforms makes it difficult to make even the most trivial changes. To compensate for this lack of flexibility, we often have to design custom, inefficient and hard-to-maintain mechanisms that add technical debt to the project.

As you've probably already guessed, we've decided to replace the current Web IDE with one built on top of VS Code. In the coming milestones, we will build out custom support for the features not already available in the VS Code core, and validate that the workflows you already depend on in the Web IDE are handled in the new experience. We're working with the team that builds our amazing GitLab Workflow extension for VS Code to make it available in the browser so we can bundle it in the Web IDE, and bring all those great features along for the ride. That includes bringing merge request comments into the Web IDE for the first time ever!

A normal Makefile for building projects with Docker would look more or less like this:

One of the main benefits from tagging your container images with the corresponding commit hash is that it's easy to trace back who a specific point in time, know how the application looked and behaved like for that specifc point in history and most importantly, blame the people that broke the application ;)

Software Bill Of Materials (SBOM) is analogous to a packing list for a shipment; it’s all the components that make up the software, or were used to build it. For container images, this includes the operating system packages that are installed (e.g.: ca-certificates) along with language specific packages that the software depends on (e.g.: log4j). The SBOM could include only some of this information or even more details, like the versions of components and where they came from.

Included in Docker Desktop 4.7.0 is a new, experimental docker sbom CLI command that displays the SBOM (Software Bill Of Materials) of any Docker image.

With every generation of computing comes a dominant new software or hardware stack that sweeps away the competition and catapults a fledgling technology into the mainstream.I call it the Canonical Stack (CS).Think the WinTel dynasty in the 80s and 90s, with Microsoft on 95% of all PCs with “Intel inside.” Think LAMP and MEAN stack. Think Amazon’s S3 becoming a near universal API for storage. Think of Kubernetes and Docker for cloud orchestration.

By offering to have a real-time technical discussion, you are reframing the code review process away from an 'instructor correcting student' dynamic and into a 'colleagues working together' mindset.

if you are on the job market looking for a team that cares more about being agile than going through the motions to look agile, ask these questions

Level 5: Stop the line. The highest level of code review comments. Borrowing the term from Toyota's manufacturing process this is when the code reviewer noticed something in the PR that signals a major defect.

Level 4: Infringement. This is where things get more serious, note that infringement means rules were broken. In this context, rules can mean a number of things, from the more obvious feature spec and framework rules to things like style guides and coding principles.

Level 3: Suggestions. These can also be thought of as recommendations and alternatives.

Level 2: Nitpicks. Usually, comments about grammar errors and minor stylistic issues/typos go there. The solution to the nitpicks is usually very obvious and if the solution is opinionated, the opinion is not strongly held. Naming a method foobarGenerator vs foobarFactory goes in here, and nitpick comments often start with Nitpick:.

Level 1: Clarifications. I also think of these as sanity checks, I may even start the comment with something like I may be completely off the base here... or Just to sanity check...

Each developer on average wastes 30 minutes before and after the meeting to context switch and the time is otherwise non-value adding. (See this study for the cost of context switching).

Overall, if speed is your primary concern and you’re on a budget, then Circle CI is the clear choice. If you’re not looking to run a ton of builds each month and your code is already in Github, then Github Actions can offer similar performance with the added convenience of having everything under one service. Even though we liked Travis better, our main criteria was value, and since you can’t use Travis for free after the first month, GitLab was able to grab the third slot, despite it being weaker in almost every other category.

Somebody once described balance to me as three buckets filled with water. One for career, a second for physical health, and a third for social and family life. At any point, one bucket might be running low. But as long as the overall water level is high enough, things should be fine.

html[theme='dark-mode'] { filter: invert(1) hue-rotate(180deg); }

Pyenv works by adding a special directory called shims in front of your PATH environment variable

If you are on Linux, you can simply download it from GitHub but the most convenient way is to use the pyenv-installer that is a simple script that will install it automatically on your distro, whatever it is, in the easiest possible way.

As of today, the Docker Engine is to be intended as an open source software for Linux, while Docker Desktop is to be intended as the freemium product of the Docker, Inc. company for Mac and Windows platforms. From Docker's product page: "Docker Desktop includes Docker Engine, Docker CLI client, Docker Build/BuildKit, Docker Compose, Docker Content Trust, Kubernetes, Docker Scan, and Credential Helper".

The diagram below tries to summarise the situation as of today, and most importantly to clarify the relationships between the various moving parts.

If you want your voice to be heard (and also improve the usability of your text) you have to design your document for “skim-ability”. You do this by providing anchor points that allow the user to gauge the content without actually reading it. You want the outline and key arguments to keep standing out in the final version of the document.

When writing an article, I generally visualize a concrete person as representative of the audience, that I am directing this text towards.

Writing is generally a great way to learn, but one has to realize that you are doing it. Learning is a slow process and requires patience. It is not helped much by agonizing in front of a screen, trying to squeeze out another sentence. Doing more research on the topic by reading a book, blog or paper and taking notes may be a better time investment.

The realization that you don’t have the complete message in your head, will often only become apparent while writing. This surfaces as inability to find a good punch-line or to express yourself clearly. In fact, writing is a great test to see if you have a good understanding of a topic, and have a firm grasp on the vocabulary of the domain.

an asshole test. You see who turns into an asshole under pressure and they don't make it to the next round".

A 20-year age difference (for example, from 20 to 40, or from 30 to 50 years old) will, on average, correspond to reading 30 WPM slower, meaning that a 50-year old user will need about 11% more time than a 30-year old user to read the same text.

Users’ age had a strong impact on their reading speed, which dropped by 1.5 WPM for each year of age.

Stop using TODO for everything in your comments that requires you to do something later.

Create the new empty table Write to both old and new table Copy data (in chunks) from old to new Validate consistency Switch reads to new table Stop writes to the old table Cleanup old table

Without accounting for what we install or add inside, the base python:3.8.6-buster weighs 882MB vs 113MB for the slim version. Of course it's at the expense of many tools such as build toolchains3 but you probably don't need them in your production image.4 Your ops teams should be happier with these lighter images: less attack surface, less code that can break, less transfer time, less disk space used, ... And our Dockerfile is still readable so it should be easy to maintain.

scratch is a special empty image with no operating system.

Overall, having spent a significant amount of time building this project, scaling it up to the size it’s at now, as well as analysing the data, the main conclusion is that it is not worth building your own solution, and investing this much time. When I first started building this project 3 years ago, I expected to learn way more surprising and interesting facts. There were some, and it’s super interesting to look through those graphs, however retrospectively, it did not justify the hundreds of hours I invested in this project.I’ll likely continue tracking my mood, as well as a few other key metrics, however will significantly reduce the amount of time I invest in it.

45% less time spent in video & audio calls that day

A properly formed Git commit subject line should always be able to complete the following sentence:If applied, this commit will your subject line hereFor example:If applied, this commit will refactor subsystem X for readability

Git itself uses the imperative whenever it creates a commit on your behalf.For example, the default message created when using git merge reads:Merge branch 'myfeature'

Commit messages with bodies are not so easy to write with the -m option. You’re better off writing the message in a proper text editor.

Firstly, not every commit requires both a subject and a body. Sometimes a single line is fine, especially when the change is so simple that no further context is necessary.

Summarize changes in around 50 characters or less More detailed explanatory text, if necessary. Wrap it to about 72 characters or so. In some contexts, the first line is treated as the subject of the commit and the rest of the text as the body. The blank line separating the summary from the body is critical (unless you omit the body entirely); various tools like `log`, `shortlog` and `rebase` can get confused if you run the two together. Explain the problem that this commit is solving. Focus on why you are making this change as opposed to how (the code explains that). Are there side effects or other unintuitive consequences of this change? Here's the place to explain them. Further paragraphs come after blank lines. - Bullet points are okay, too - Typically a hyphen or asterisk is used for the bullet, preceded by a single space, with blank lines in between, but conventions vary here If you use an issue tracker, put references to them at the bottom, like this: Resolves: #123 See also: #456, #789

Separate subject from body with a blank lineLimit the subject line to 50 charactersCapitalize the subject lineDo not end the subject line with a periodUse the imperative mood in the subject lineWrap the body at 72 charactersUse the body to explain what and why vs. how

What may be a hassle at first soon becomes habit, and eventually a source of pride and productivity for all involved.

Commit messages can do exactly that and as a result, a commit message shows whether a developer is a good collaborator.

A diff will tell you what changed, but only the commit message can properly tell you why.

Look at Spring Boot, or any repository managed by Tim Pope.

To access your Linux files in Windows, open the Ubuntu terminal and type explorer.exe . (include the punctuation mark). This will open the linux directory in Windows Explorer, with the WSL prefix “\wsl$\Ubuntu-18.04\home\your-username”.Now, you’ll notice that Windows treats your Linux environment as a second network.

The eagle-eyed among you may notice that there isn't a .git directory in the app-example-2 working tree. Instead, there's a .git file. This file points to the git directory in the original clone, and it means that all your usual git commands work inside the app-example-2 directory, as well as in the original app-example directory.

Anecdotally, I've found IDEs get much less "confused" if you use their built-in support for switching git branches, instead of changing them from the command line and waiting for the IDE to "notice" the changes.

Reading puts candidates at ease compared to writing code.  As an interviewer, stress is your enemy because it raises adrenaline which lowers IQ by several points, causing you to miss good candidates.   Candidates prefer reading partly because they are relieved to not have to write code, but also because the interviewer can easily adjust the reading questions to accommodate for the candidate’s skill.

Finally, to make our terminal really pretty, we need to customize the prompt. There's lots of options out there for this, but the most popular one seems to be ohmyzsh for Bash and oh-my-posh for PowerShell. I'm not a huge fan of these because in my experience they slow down the terminal to a point which makes me frustrated to use them, and since they are separate solutions for each environment they must be configured separately.

We asked them to collect searches they performed in their normal workday, and then evaluate their performance on Google, Bing, Neeva, You.com, Kagi, and DuckDuckGo.Here are more examples of programming queries where Google is lagging behind.

In the previous version, using the standard library, once the data is loaded we no longer to keep the file open. With this API the file has to stay open because the JSON parser is reading from the file on demand, as we iterate over the records.

One common solution is streaming parsing, aka lazy parsing, iterative parsing, or chunked processing.

Then, if the string can be represented as ASCII, only one byte of memory is used per character. If the string uses more extended characters, it might end up using as many as 4 bytes per character. We can see how much memory an object needs using sys.getsizeof()

I tried PuTTY, which I’d heard was the good SSH thing on Windows, but it’s…​ not good, at all. PowerShell does come with an SSH client, so once you have this working with a reasonable terminal, you can use SSH as normal.

I sometimes wondered why the VS Code team put so much effort into the built-in terminal inside the editor. I tried it once on Linux and never touched it again, because the terminal window I had right next to my editor was just massively better in every way. Having used Windows terminals for a while, I now fully understand why it’s there.

Scoop seems to be the best of the bunch, so far?

They just automate the process of going to the website, downloading an installer and then running it - which is slightly better than doing it yourself.

It’s got less useful stuff in it than most Linux distro “app stores” and is utterly miniscule compared to the Debian repositories, which have ~60,000 packages in, or Arch’s AUR, with 73,000 (these counts include the whole Linux OS, though, with is installed using the same package manager).

if you want to add persistent environment variables to your currently running shell, you should put a setx command in your $profile file and then reload it: . $profile - or maybe run myvar="value" && setx %myvar% "value", or something similar.

C:\Users\DuncanLock\Documents\PowerShell\Microsoft.PowerShell_profile.ps1~/.bashrc

PowerShell does have a ~ alias for your home folder, and cd ~ works!

You can get Windows versions of most of the standard *nix userland utils, which seem to work OK with PowerShell

Desktop Linux is often criticized for this, but Windows is much worse, somehow! It’s really inconsistent. Half of it is “new” UI and half of it is old Win32/GDI type UI - just as bad as KDE/GTK - except worse, because you can’t configure them to use the same theme. Also, when you install a Linux distribution, it’ll start off either all KDE or all GTK, or whatever - but with Windows you’re stuck with a random mix of both right from the start.

Most companies are not prepared to pay for a staging environment identical to production

The Code Review Pyramid

With personal stability (job, hobbies, network, etc.) and age you care less and less what people think and put less emphasis on what people think about you. - At the same time, you judge people less and less because, what does your judgment really matter? And when you judge people less you judge yourself less. - A lot of awkwardness is a function of trying hard to fit a social shape and failing. When you let go of most social expectations of yourself and others, in my experience mostly because of age and stability, you lose the awkwardness.

Lists promise value quickly A normal essay promises a payoff only when you finish it. But lists promise at least some value almost immediately.

Let’s look at a recent paper by Xia, Bao, Lo, Xing, Hassan, & Li entitled Measuring Program Comprehension: A Large-Scale Field Study with Professionals and published in IEEE Transactions on Software Engineering, 44, 951-976, 2018. This paper is quite interesting in that it describes in great details how the figures are obtained. And it says that Comprehension took on average ~58%.

# Input Input: 123, Output: Input: 121, Output: Input: 111, Output: Input: 123454321, Output: Input 123123, Output: # Instruction Output true if input is a palindrome # Output Input: 123, Output: false Input: 121, Output: true Input: 111, Output: true Input: 123454321, Output: true Input 123123, Output: false

Using named arguments is nice for languages that support it, but this is not always a possibility. Even in Python, where time.sleep is defined with a single argument named secs, we can’t call sleep(secs=300) due to implementation reasons. In that case, we can give the value a name instead.Instead of this:time.sleep(300)Do this:sleep_seconds = 300 time.sleep(sleep_seconds)Now the code is unambiguous, and readable without having to consult the documentation.

defaults write NSGlobalDomain WebKitDeveloperExtras -bool true defaults write -g WebKitDeveloperExtras -bool YES

Core: like an engine in a car. The product is meaningless without it.Necessary: like a car’s spare wheel. It’s rarely used but when needed, its function decides the success of the system.Added value: like a car’s cup-holder. It’s nice to have but the product is perfectly usable without it.Unique Selling Point: the main reason people should buy your product instead of your rivals. For example, your car is the best off-road vehicle.

Here is an architecture diagram of the devices I’m currently using

Have you ever built an image only to realize that you actually need it on a user account other than root, requiring you to rebuild the image again in rootless mode? Or have you built an image on one machine but run containers on the image using multiple different machines? Now you need to set up an account on a registry, push the image to the registry, Secure Shell (SSH) to each device you want to run the image on, and then pull the image. The podman image scp command solves both of these annoying scenarios as quickly as they occur.

an onion address is a promise and a mechanism to assure that you are taking seriously the needs of the people who use Tor.

As mentioned earlier, PATCH requests should apply partial updates to a resource, whereas PUT replaces an existing resource entirely. It's usually a good idea to design updates around PATCH requests

Aside from using HTTP status codes that indicate the outcome of the request (success or error), when returning errors, always use a standardized error response that includes more detailed information on what went wrong.

https://api.averagecompany.com/v1/health https://api.averagecompany.com/health?api_version=1.0

When dealing with date and time, APIs should always return ISO 8601-formatted strings.

So, to summarize, don't ask "Any Java experts around?", but rather ask "How do I do [problem] with Java and [other relevant info]?"

You're asking people to take responsibility. You're questioning people's confidence in their abilities. You're also unnecessarily walling other people out. I often answer questions related to languages or libraries I have never used, because the answers are (in a programmer kind of way) common sense.

This isn’t MAIN_PRIORITY, so we aren’t going to do it until at least ESTIMATED_DONE_DATE. Right now our priority is MAIN_PRIORITY because of ONE_SENTENCE_JUSTIFICATION, and this is 100% our shipping focus. I agree this sounds like a really useful feature - once we finish MAIN_PRIORITY, should we consider dropping SECOND_PRIORITY and do it?

for debugging purposes, a good combination is --lf --trace which would start a debug session with pdb at the beginning of the last test that failed:

pytest -l

If you start pytest with --pdb, it will start a pdb debugging session right after an exception is raised in your test. Most of the time this is not particularly useful as you might want to inspect each line of code before the raised exception.

pytest --lf

pytest -x

pytest --collect-only

pytest test_validator.py::test_regular_email_validates

Apart from shared fixtures you could place external hooks and plugins or modifiers for the PATH used by pytest to discover tests and implementation code.

pytest can read its project-specific configuration from one of these files: pytest.ini tox.ini setup.cfg

To have the fixture actually be used by one of your test, you simply add the fixture’s name as an argument

But the problem with Poetry is arguably down to the way Docker’s build works: Dockerfiles are essentially glorified shell scripts, and the build system semantic units are files and complete command runs. There is no way in a normal Docker build to access the actually relevant semantic information: in a better build system, you’d only re-install the changed dependencies, not reinstall all dependencies anytime the list changed. Hopefully someday a better build system will eventually replace the Docker default. Until then, it’s square pegs into round holes.

Third, you can use poetry-dynamic-versioning, a plug-in for Poetry that uses Git tags instead of pyproject.toml to set your application’s version. That way you won’t have to edit pyproject.toml to update the version. This seems appealing until you realize you now need to copy .git into your Docker build, which has its own downsides, like larger images unless you’re using multi-stage builds.

But if you’re doing some sort of continuous deployment process where you’re continuously updating the version field, your Docker builds are going to be slow.

VCR.py works primarily via the @vcr decorator. You can import this decorator by writing: import vcr.

The VCR.py library records the responses from HTTP requests made within your unit tests. The first time you run your tests using VCR.py is like any previous run. But the after VCR.py has had the chance to run once and record, all subsequent tests are:Fast! No more waiting for slow HTTP requests and responses in your tests.Deterministic. Every test is repeatable since they run off of previously recorded responses.Offline-capable! Every test can now run offline.

 75% of people in the U.S. never tweet.On an average weeknight in January, just 1% of U.S. adults watched primetime Fox News (2.2 million). 0.5% tuned into MSNBC (1.15 million).Nearly three times more Americans (56%) donated to charities during the pandemic than typically give money to politicians and parties (21%).

The ENTRYPOINT specifies a command that will always be executed when the container starts. The CMD specifies arguments that will be fed to the ENTRYPOINT.

DevOps is an interesting case study for understanding MLOps for a number of reasons: It underscores the long period of transformation required for enterprise adoption.It shows how the movement is comprised of both tooling advances as well as shifts in cultural mindset at organizations. Both must march forward hand-in-hand.It highlights the emerging need for practitioners with cross-functional skills and expertise. Silos be damned.

MLOps today is in a very messy state with regards to tooling, practices, and standards. However, this is to be expected given that we are still in the early phases of broader enterprise machine learning adoption. As this transformation continues over the coming years, expect the dust to settle while ML-driven value becomes more widespread.

LXC, is a serious contender to virtual machines. So, if you are developing a Linux application or working with servers, and need a real Linux environment, LXC should be your go-to. Docker is a complete solution to distribute applications and is particularly loved by developers. Docker solved the local developer configuration tantrum and became a key component in the CI/CD pipeline because it provides isolation between the workload and reproducible environment.

# line containing 'cake' but not 'at' # same as: grep 'cake' table.txt | grep -v 'at' # with PCRE: grep -P '^(?!.*at).*cake' table.txt $ awk '/cake/ && !/at/' table.txt blue cake mug shirt -7

Jeżeli masz dylemat czy użyć CMD, czy ENTRYPOINT jako punkt startowy twojego kontenera, odpowiedz sobie na następujące pytanie.Czy zawsze moje polecenie MUSI się wykonać? Jeśli odpowiedź brzmi tak, użyj ENTRYPOINT. Co więcej, jeśli potrzebujesz przekazać dodatkowe parametry, które mogą być nadpisane podczas uruchomienia kontenera — użyj również instrukcji CMD.

More backlinks to you and your work: Being the teammate that contributes to the system of knowledge shared shows how much you care about the success of the organization. And, it does it in a way that helps you have more documented and attributable credibility for the value you create within your organization.

Writing supplants meetings: When there is good documentation around a meeting (briefs, meeting notes, etc.), meetings can be leaner and more productive because people don’t have to be in the room to know what’s happening. So, only those who are actively contributing to the discussion need to attend.

Without meeting notes and documentation, companies become reliant on unreliable verbal accounts, 1:1 updates, and needing to be in the room to get things done.

Scales everyone’s knowledge: think about how many people you interact with on a given work day. I’m talking about the real human to human kind where you relay your ideas to others. It’s probably in the 5-20 range. If you put those same ideas into a doc or an email, your distribution goes to infinity. Anyone can read it.

== and != for string comparison -eq, -ne, -gt, -lt, -le -ge for numerical comparison

> will overwrite the current contents of the file, if the file already exists. If you want to append lines instead, use >>

The syntax for “redirecting” some output to stderr is >&2. > means “pipe stdout into” whatever is on the right, which could be a file, etc., and &2 is a reference to “file descriptor #2” which is stderr.

single quotes, which don’t expand variables

This only works if you happen to have Bash installed at /bin/bash. Depending on the operating system and distribution of the person running your script, that might not necessarily be true! It’s better to use env, a program that finds an executable on the user’s PATH and runs it.

It isn't about note taking, it's about supporting a brain that simply isn't capable of retaining the level of information that we have to deal with. The ultimate note taking device is actually an augmented human brain that has perfect recollection and organisation.

In Japan, the country where the first emoji sets originated, red is traditionally used to represent increases in the value of a stock. Meanwhile, green is used to represent decreases in stock value.

the curse of knowledge. It’s a simple but devastating effect: Once we know something, it’s very difficult to imagine not knowing it, or to take the perspective of someone who doesn't.

Adopting Kubernetes-native environments ensures true portability for the hybrid cloud. However, we also need a Kubernetes-native framework to provide the "glue" for applications to seamlessly integrate with Kubernetes and its services. Without application portability, the hybrid cloud is relegated to an environment-only benefit. That framework is Quarkus.

Kubernetes-native is a specialization of cloud-native, and not divorced from what cloud native defines. Whereas a cloud-native application is intended for the cloud, a Kubernetes-native application is designed and built for Kubernetes.

According to Wilder, a cloud-native application is any application that was architected to take full advantage of cloud platforms. These applications: Use cloud platform services. Scale horizontally. Scale automatically, using proactive and reactive actions. Handle node and transient failures without degrading. Feature non-blocking asynchronous communication in a loosely coupled architecture.

two main problems with framing decisions and policies in terms of usefulness: (1) being useful is not always to our own benefit – sometimes, we are being used as a means to someone else’s end, and we end up miserable as a result; and (2) the lenses themselves of usefulness and uselessness can obscure our view of the good life.

The script is in batch with some portions of powershell. The base code is fairly simple and most of it came from Googling ".bat transfer files" followed by ".bat how to only transfer certain file types" etc. The trick was making it work with my office, knowing where to scan for new files, knowing where not to scan due to lag (seriously, if you have a folder with 200,000 .txt files that crap will severally slow down your scans. Better to move it manually and then change the script to omit that folder from future searches)

It essentially scans the on-site drive for any new files, generates hash values for them, transfers them to the Cloud, then generates hash values again for fidelity (in court you have to prove digital evidence hasn't been tampered with).

Salesforce has a unique use case where they need to serve 100K-500K models because the Salesforce Einstein product builds models for every customer. Their system serves multiple models in each ML serving framework container. To avoid the noisy neighbor problem and prevent some containers from taking significantly more load than others, they use shuffle sharding [8] to assign models to containers. I won’t go into the details and I recommend watching their excellent presentation in [3].

Batching predictions can be especially beneficial when running neural networks on GPUs since batching takes better advantage of the hardware.

Inference Service — provides the serving API. Clients can send requests to different routes to get predictions from different models. The Inference Service unifies serving logic across models and provides easier interaction with other internal services. As a result, data scientists don’t need to take on those concerns. Also, the Inference Service calls out to ML serving containers to obtain model predictions. That way, the Inference Service can focus on I/O-bound operations while the model serving frameworks focus on compute-bound operations. Each set of services can be scaled independently based on their unique performance characteristics.

Provide a model config file with the model’s input features, the model location, what it needs to run (like a reference to a Docker image), CPU & memory requests, and other relevant information.

what changes when you need to deploy hundreds to thousands of online models? The TLDR: much more automation and standardization.

“Shadow Mode” or “Dark Launch” as Google calls it is a technique where production traffic and data is run through a newly deployed version of a service or machine learning model, without that service or model actually returning the response or prediction to customers/other systems. Instead, the old version of the service or model continues to serve responses or predictions, and the new version’s results are merely captured and stored for analysis.

you can also mount different FastAPI applications within the FastAPI application. This would mean that every sub-FastAPI application would have its docs, would run independent of other applications, and will handle its path-specific requests. To mount this, simply create a master application and sub-application file. Now, import the app object from the sub-application file to the master application file and pass this object directly to the mount function of the master application object.

There are officially 5 types of UUID values, version 1 to 5, but the most common are: time-based (version 1 or version 2) and purely random (version 3). The time-based UUIDs encode the number of 10ns since January 1st, 1970 in 7.5 bytes (60 bits), which is split in a “time-low”-“time-mid”-“time-hi” fashion. The missing 4 bits is the version number used as a prefix to the time-hi field.  This yields the 64 bits of the first 3 groups. The last 2 groups are the clock sequence, a value incremented every time the clock is modified and a host unique identifier.

Even though most posts are warning people against the use of UUIDs, they are still very popular. This popularity comes from the fact that these values can easily be generated by remote devices, with a very low probability of collision.

This basic example compiles a simple Go program. The naive way on the left results in a 961 MB image. When using a multi-stage build, we copy just the compiled binary which results in a 7 MB image.

# Image size: 7 MB

FROM golang:1.17.5 as builder

WORKDIR /workspace
COPY . .
ENV CGO_ENABLED=0
RUN go get && go build -o main .

FROM scratch
WORKDIR /workspace
COPY --from=builder \
     /workspace/main \
     /workspace/main

CMD ["/workspace/main"]

Docker introduced multi-stage builds starting from Docker Engine v17.05. This allows us to perform all preparations steps as before, but then copy only the essential files or output from these steps.

Making a small change to a file or moving it will create an entire copy of the file. Deleting a file will only hide it from the final image, but it will still exist in its original layer, taking up space. This is all a result of how images are structured as a series of read-only layers. This provides reusability of layers and efficiencies with regards to how images are stored and executed. But this also means we need to be aware of the underlying structure and take it into account when we create our Dockerfile.

In this example, we created 3 copies of our file throughout different layers of the image. Despite removing the file in the last layer, the image still contains the file in other layers which contributes to the overall size of the image.

FROM debian:bullseye

COPY somefile.txt . #1

# Small change but entire file is copied
RUN echo "more data" >> somefile.txt #2

# File moved but layer now contains an entire copy of the file
RUN mv somefile.txt somefile2.txt #3

# File won't exist in this layer,
# but it still takes up space in the previous ones.
RUN rm somefile2.txt

We’re just chmod'ing an existing file, but Docker can’t change the file in its original layer, so that results in a new layer where the file is copied in its entirety with the new permissions.In newer versions of Docker, this can now be written as the following to avoid this issue using Docker’s BuildKit:

FROM debian:bullseye

COPY somefile.txt .

RUN chmod 777 somefile.txt

FROM debian:bullseye

COPY --chmod=777 somefile.txt .

when you make changes to files that come from previous layers, they’re copied into the new layer you’re creating.

Many processes will create temporary files, caches, and other files that have no benefit to your specific use case. For example, running apt-get update will update internal files that you don’t need to persist because you’ve already installed all the packages you need. So we can add rm -rf /var/lib/apt/lists/* as part of the same layer to remove those (removing them with a separate RUN will keep them in the original layer, see “Avoid duplicating files”). Docker recognize this is an issue and went as far as adding apt-get clean automatically for their official Debian and Ubuntu images.

An important way to ensure you’re not bringing in unintended files is to define a .dockerignore file.

# Ignore git and caches
.git
.cache

# Ignore logs
logs

# Ignore secrets
.env

# Ignore installed dependencies
node_modules

...

You can save any local image as a tar archive and then inspect its contents.

bash-3.2$ docker save <image-digest> -o image.tar
bash-3.2$ tar -xf image.tar -C image
bash-3.2$ cd image
bash-3.2$ tar -xf <layer-digest>/layer.tar
bash-3.2$ ls

Technology → Cool Stuff to Work OnIntellection → Smart People to Work WithCertainty → Repeatability in Work Environment

Smalltalk image is unlike a collection of Java class files in that it can store your programs (both source and compiled bytecode), their data, and execution state. You can quit and save your work while some code is executing, move your image to an entirely different machine, load your image… and your program picks up where it left off.

Smalltalk has a virtual machine which allows you to execute your code on any platform where the VM can run. The image however is not only all your code, but the entire Smalltalk system, including said virtual machine (because of course it’s written in itself).

The most important thing you can do is provide actionable feedback. This means being specific about your observations of their work. It also means providing direction about what they could have done differently or what they need to learn or practice in order to improve. If your mentee doesn’t know what to do next, your feedback wasn’t actionable.

So when the morale on the team dips (usually related to lack of clarity or setbacks making things take a long time) people leave the team, which further damages morale and can easily result in a mass exodus.

People come and go (which has the net impact of making you feel like you are just a resource)

Instead of “I have a type, it’s called MyType, it has a constructor, in the constructor I assign the property ‘A’ to the parameter ‘A’ (and so on)”, you say “I have a type, it’s called MyType, it has an attribute called a”

attrs lets you declare the fields on your class, along with lots of potentially interesting metadata about them, and then get that metadata back out.

>>> Point3D(1, 2, 3) == Point3D(1, 2, 3)

    def __eq__(self, other):
        if not isinstance(other, self.__class__):
            return NotImplemented
        return (self.x, self.y, self.z) == (other.x, other.y, other.z)
    def __lt__(self, other):
        if not isinstance(other, self.__class__):
            return NotImplemented
        return (self.x, self.y, self.z) < (other.x, other.y, other.z)

>>> Point3D(1, 2, 3)

def __repr__(self):
    return (self.__class__.__name__ +
        ("(x={}, y={}, z={})".format(self.x, self.y, self.z)))

Look, no inheritance! By using a class decorator, Point3D remains a Plain Old Python Class (albeit with some helpful double-underscore methods tacked on, as we’ll see momentarily).

class Point3D(object):
    def __init__(self, x, y, z):
        self.x = x
        self.y = y
        self.z = z

import attr
@attr.s
class Point3D(object):
    x = attr.ib()
    y = attr.ib()
    z = attr.ib()