The TMPDIR podcast

By TMPDIRTechnology

The TMPDIR podcast episodes

December 02, 2022Databases
Available on your favorite podcast platform.
In this episode, we are joined by Blake Miner
to discuss databases.
Show notes:
Introduction
Blake's background, what type of work he does, company information, etc.
Data has substantial value
What is a database?
General overview - store data, process it, and look it up later
Important that data stays intact in the event of a app/system crash, power
loss, etc.
Important that conflicting concurrent changes are handled appropriately
What safety guarantees do I get? -> Transactions / ACID - Atomicity,
Consistency, Isolation, Durability
atomicity -> abortability guarantee for a group of writes (i.e. upon
failure)
consistency -> guarantee that certain (app-specific; sometimes
database-specific) constraints are not violated
isolation -> describes how database handles concurrent edits of the same
data. Serializable is the strongest guarantee. Snapshot isolation is very
common.
durability -> storage guarantee (i.e. upon commit, data has fsync'ed to
disk). Study: SSDs get 1 bad block in first 4 years; disaster could wipe
out an entire data center; nothing is perfect.
Storage / lookup: Data models / data structures
Derived data; caching
Concurrency; consistency & isolation
Reactivity/Realtime
tradeoffs between reacting to DB changes vs using a message bus
streaming platforms such as NATS Jetstream and Kafka
SQL
Tables and rows -> array of structs
SQL -> Declarative query language (query engine determines implementation
details at runtime)
B-tree Indexes
NoSQL Databases
"NoSQL" -> a marketing term
Redis
Data structure store
In-memory vs. disk-based
Embedded databases (i.e. API vs. commands over TCP)
Designing Data-Intensive Applications by Martin
Kleppmann. Blake recommends this book.
Discuss this episode at our
community site.
1 hr 4 min
January 07, 2022Do you own your platform?
Available on your favorite podcast platform.
Show notes:
difference between a product and platform
platform is a collection of modules and tools
a platform is extensible, programmable, consists of reusable pieces and
components
desirable qualities in a platform and various considerations
a platform is a living being to some extent
availability
stability
performance
how easy is it to fix when things break
adaptability
the iceberg analogy -- most of your product is invisible, but this can cause
problems
how do you select the right technology for your platform?
important not have bias
products are short-lived, where platforms last a long time, so important to
make the right choices in your platform.
it is good to depend on standards to avoid lock-in to one tool/technology
the wrong solutions do not scale
have a network of experts you can reach out to and get the information you
need
knowing who to ask is the superpower in this age
why you need some level of control over your platform
damage control: solve problems when they arise
moving things forward: can it be modified as new requirements come up
building platforms for reuse
build systems are important
automation
continuous integration
The Yoe Distribution and the Simple IoT project are examples of platforms that
are re-usable and flexible, and applicable for a wide range of projects
recomposable platforms scale not just vertically, but horizontally, which is
the real power of them
a lot of up-front discipline goes into this
you create a platform not only for others, but yourself
you respect the APIs in the same way you expect others to respect them
APIs as a contract become very important
Most people can track product costs very well. Platform costs are harder to
track and are an investment in the future
example of an embedded Linux platform that goes into a product that lives
for many years. A lot of work is required upfront to get the base system
running, but then you can build on it for years.
A more constrained platform like a MCU + RTOS tends to be not as extensible
and more limited to a product.
Two kinds of platforms
Transaction platforms: facebook, twitter, etc
Innovation platforms: what we've been talking about
personal platform
write things down
when helping someone, instead of emailing or messaging the solution, I write
down the solution in some type of long lived documentation, and then share
that documentation with them.
do you own your platform?
owning your platform is more of a mindset rather than an rigid absolute
are you willing to learn and own the details of your platform
with OSS we not only use it, but we also have the option to be part of it
Discuss this episode at our
community site.
33 min
December 31, 2021 A day in the life of a Go developer
In this episode, Cliff describes how he develops Go code. We cover tooling,
testing, deployment, and all the things that make Go development a great
experience.
Why are Go applications so reliable?
Available on your favorite podcast platform.
Discuss this episode at our
community site.
25 min
May 21, 2021A day in the life of a Yoe developer
In this episode, Khem describes how he develops, tests the Yoe distribution.
Show notes:
what does your day to day development look like
recent platforms developed with Yoe
why we mirror repositories (note, these are not forks)
we use Yoe to test combinations of things
Yoe testing strategy and tools
master branch is usually stable
how we generate the changelog and why it might be useful for any Yocto
developer
Available on your favorite podcast platform.
Discuss this episode at our
community site.
25 min
April 15, 2021Discussion with Andrew Wafaa
Our guest on this episode is
Andrew Wafaa. We discussed many
aspects of open source and technology. Near the end of the episode, Andrew
shares some advice for people who want to learn about technology.
Show notes:
Andrew has been involved in OSS for 2 decades
likes packaged software -- started by packaging software for SUSE for his own
use
too many people are concerned about what others will think of their work
you keep doing OSS for yourself.
successful community requires a safe, open, collaborative environment
accept the fact you can't please everyone. Once you try to please everyone,
you please no one.
marketing is the Achilles heel of many projects
events are an easy way to market projects
BSD
very nice community, and BSP projects seem to be thriving
collaboration between commercial entities is better than Linux -- which may
seem seem a bit surprising considering BSD is licensed with a permissive
license
used a lot in education -- great learning platform
tips
read the documentation
build up a portfolio (Github, etc)
try it all out
don't be afraid to ask
have fun
operating systems/distros are a good way to get exposed to a lot of
different technologies -- languages, compilers, projects, etc.
Available on your favorite podcast platform.
Discuss this episode at our
community site.
54 min
April 02, 2021Documentation
In this episode we discus:
What is documentation
Benefits of documentation
Documentation is critical for scaling, collaboration, and distributed teams
How documentation benefits ourselves
Lesli Lamport's thoughts: To think, you have to write. If you’re thinking
without writing, you only think you’re thinking.
https://www.microsoft.com/en-us/research/wp-content/uploads/2016/07/leslie_lamport.pdf
Code should not only be created for the compiler, but more importantly for
other people
Importance of idioms, automatic code formatting, and code linters
Documentation should live close to the code (embedded in the code, or Markdown
in the same repository)
Benefits of Markdown
Trade-offs between Markdown and other formats like Word
How to optimize documentation for collaboration
Biggest problem with documentation is that most of it gets stale and how to
prevent this
Documentation can be a platform for thinking about a design
Discussion should happen around documentation, not the other way around
Use CI tools to publish your documentation
Useful tools: draw.io and
Typora
The futility of emailing documents
Storing documents in Git and using pull requests (PRs) is a hugely beneficial
for collaboration
Cal Newport's new book: A world without email
https://www.calnewport.com/books/a-world-without-email/
Gitlab's handbook has a lot of good ideas
https://about.gitlab.com/handbook/
The more experienced you are, the more should you document
Tools for generating documentation web sites
Docusaurus
Docsy (Hugo)
Pandoc
mdBook
mkdocs
Zola
book theme
GitBook
Example of collecting documentation from multiple projects and creating a
documentation site using Zola: https://github.com/simpleiot/docs.simpleiot.org
The TMPDIR handbook:
https://gitlab.com/tmpdir/handbook
Jupyter Notebook
https://jupyter.org/
Treat documents as code, and code as documents!
Available on your favorite podcast platform.
Discuss this episode at our
community site.
41 min
February 26, 2021Tips for Yocto BSPs
In this episode we discus:
what is a Yocto BSP
who are a BSP's users
the purpose of a BSP
examples of good BSPs
importance of keeping BSPs and Distros separate
dynamic layers
issues with forking major components like GStreamer
and many other aspects of BSPs
Available on your favorite podcast platform.
Discuss this episode at our
community site.
46 min
February 12, 2021Using Elm for frontend development
In this episode we discus:
does Elm really live up to the expectation of no run time exceptions?
Elm tooling
elm-ui
how Elm interacts with Javascript
how Elm compares to other front end technologies. As a follow up we
documented bundle size for various frameworks.
why Elm is a really good option for those of us who are not front end experts.
Available on your favorite podcast platform.
Bonus:
What are your favorite things about Elm?
Discuss this episode at our
community site.
31 min
January 22, 2021The Yoe Updater
In this episode we discus:
review of various Linux update technologies
discussion of experiments using swupdate
description of the Yoe updater
design ideas behind the Yoe updater
recent work done on the Yoe updater to make it more generic
Yoe handles installation, system initialization as well as update - this is
very critical
designed to support products with long life cycles where things may change
(partition layout, etc.)
how to add Yoe updater support to a new platform
what pieces of the system does the Yoe updater touch
how the Yoe updater handles rootfs corruption
why system initialization is important for reliable systems
why system initialization make sense in an initramfs
keep it simple is really the mantra
how the Yoe updater helps developers streamline their testing process by
removing friction
future plans of managing updates using Simple IoT
See also the
documentation
and
implementation.
Available on your favorite podcast platform.
Discuss this episode at our
community site.
47 min
January 11, 2021Reflections on developing an Odroid-C2 based product
In this episode we discus:
hardware options for implementing Embedded Linux systems (SOM, SBC, full
custom)
why we selected the Odroid-C2 for this project
review of the BSP support for the Odroid-C2
make sure you are worked with supported technology
description of what it takes to use a watchdog to protect the boot process
discussions on the trade-offs with the update architecture (we used a dual
rootfs method on this project)
trade-offs with implementing a read-only rootfs
application architecture
written in Go
local display using a Qt WebEngine to implement a full screen browser on LCD
used InfluxDB for storing time series data
used Grafana to view engineering data
how we do builds, release/changelog, and versioning
options for browser implementations
Available on your favorite podcast platform.
Discuss this episode at our
community site.
44 min

About The TMPDIR podcast

From the publisher's feed

TMPDIR is a podcast about embedded Linux, Yocto/OpenEmbedded, Zephyr, and IoT product engineering. Cliff Brake and Khem Raj dig into BSPs, OTA updates, Linux kernel and device tree work, firmware and…