August 2020

313 tweets

Replying to @Grady_Booch

@Grady_Booch
@petkasp I think you made it very clear that you
believe that only a certain way of education will produce "good" developers. And
that it can't be in done through a streamlined program.

I don't agree with that, because many skills that "good" developers have are
nevertheless not taught.

Replying to @coderbyheart

@Grady_Booch
@petkasp I believe that a well designed
streamlined curriculum can give students great starting points for their own
further education but will enable them to start earning a full salary in shorter
time, which will allow them to have more ease of mind to study topics they want
to explore.

Replying to @Grady_Booch

@Grady_Booch These are median salaries from
Burning Glass, and I assume for the entire U.S. So that salary might very well
be decent for entry level positions if you do not live in the Valley.

Especially if you can earn money 3-4 years earlier with much less student debt.

#e2e test for @bifravst on
@Azure are getting somewhere ... however every day
brings new hurdles. Today I discovered that I can't programmatically
authenticate users against Azure AD B2C using the same function app, because it
can only have one Issuer Url configured for auth.
Embedded Photo

Replying to @coderbyheart

So this is fine for CI #e2e test, but you it means you can't easily run these
test against a development instance. :/

Also, the Issuer URL has a different value for developer generated tokens. Which
means a different config in prod vs. test. 🤕

Replying to @omakoleg

@omakoleg Why is that so bad? As you wrote,
there are different entry points, so the folder may be the same but the handler
is different.

Of course it would be cleaner to extract shared code into a library and group
lambdas per "feature" of your solution.

Replying to @Niklas_L

@Niklas_L I think ideally it should still accept
requests (204), and queue them up (e.g. if they are write requests and can't be
handled immediately). If you don't have a strict contract on application of
writes, of course.

Replying to @Niklas_L

@Niklas_L So you mean in the case where the HA
proxy thinks the target is fine, forwards the request and it fails? That would
result in a timeout, and then the HA proxy can retry either the same host, or
another in the resource group.

@thenativeweb Das schwierige an #Architektur
ist ja auch oft zu erkennen, was eine gute Architektur aus macht, also warum
wir uns überhaupt damit auseinandersetzen sollten. Woran ist zu erkennen, dass
eine Architektur besser oder schlechter durch den aktuellen change wird?

Replying to @danvacanti

@danvacanti
@prokanban Maybe encountering yet another
certification of theoretical knowledge about practical competences with bogus
value proposition ("industry recognized PK I Certification") justifies a GIF?
You might consider taking this as constructive feedback on how you market your
certification.

Progress in the right direction: #e2e tests now running for
@bifravst on
@github actions against @azure.

I am still not able to test that setting up the entire solution works from a
blank account, but at least I can verify that updating an existing one does not
break it.
Embedded Photo

Virtually no project management tool has a feature connecting issues to goals,
or visualizes hypothesized influence on key metrics.

The reason you should work on an old issue in a issue tracker should not be
because it's in there but because the goal it supports is still valid.

Replying to @chrmaske

@chrmaske Yes, but these are addons to the
regular way to structure work, and often hidden down in the visual/workflow
hierarchy issues are presented. This needs to be up front and center: you should
not be allowed to start working on issues without going through a goal selection
first.

@github While developing actions in PRs, they
often fail, and after I push a change I have to navigate around to find the
latest execution. It would be great if the Action UI would get a push
notification about a newer execution for the same job and I could click a link
to there.

:/ trying to get rid of some flaky @Azure IoT end
to end tests for @bifravst: sometimes the
connection from the GitHub Actions runner is just reset, and somehow this also
crashes the connection using mqtt.js 4.2.0 so it can't catch and retry it
gracefully in Node.
Embedded Photo

Replying to @TonyBologni

@TonyBologni Yeah, it's much better here in
Norway, I know really cool developers working in the public sector. The salary
is good, and what you can earn on the free market as a consultant is not
significantly higher.

Replying to @mahoriR

@mahoriR
@GergelyOrosz I don't think the sentiment
was that frontend is not complex or does not provide a challenging career path.

Instead OP wanted to point out that backend heavy teams provide a great
opportunity, which might not be there if you stay in the frontend track in these
teams.

Replying to @coderbyheart

It's gut wrenching to read through this and thank you for sharing this in such a
detail @Methystic. I am however not surprised.
Far too few men in the position to make a change are taking on their
responsibility to do so. They have no incentive in the short run to do so.

Replying to @coderbyheart

It will be new organizations, built around different values that will force them
to change eventually. But then it will be too late. You might be better off
looking at the new, not at the old.

I've been using @github daily for many years,
still this UI element confuses me every time I use it.

The notification button:

  • icon shows current state
  • text indicates the action to take
  • dropdown shows no option that matches text

Looking a little bit into the future of browser apps which will no longer use
pre-compiled code using e.g. #webpack, but fetch dependencies using #esm
(ECMAScript module): we need to shift testing environments to actually use
browsers (albeit headless) as the test runner.

Replying to @coderbyheart

And what about our beloved package-lock.json? It guarantees that all
dependencies and sub-dependencies are installed with a specific version, which
ensures that everyone building and executing the code gets a specific state (see
@ReproBuilds). #ESM only pins the top-level.

Replying to @coderbyheart

It's a classical IT development: progress on one axis (improves developer UX,
decreases JS payload size and time to interaction on user side) but at the same
time creates new challenges (or resurfaces old).

Replying to @mxstbr

@mxstbr @GatsbyJS
Thank you for demonstrating that you and others are aware of what is wrong. It
would be interesting to see what your questions and proposed actions are, but I
can understand that it might be sensible to give leadership a while to prepare
an answer before making those public.

Replying to @Donorbox

@Donorbox I understand that I don't need to use
Zapier. But to charge US$17/month for being able to call {GET} /api/v1/campaigns
is a bit strange, given that you already charge 1.5% transaction fee. I would
appreciate a pay-per-request scheme, instead.

Replying to @BLJENGINEERING

@BLJENGINEERING
@NordicTweets It's quite hard to brick them,
really. I do flashing quite a lot and it never happened to me. Plug it into USB
and check what's printed on the console. No lights often means it's stuck in a
reboot loop because the flashing failed. Just reflashing the FW will solve this
typically.

Replying to @BLJENGINEERING

I've upgraded my old desk (it was used in the last century in my hometown to
store punch cards in a shoe manufactury) to a standing desk. I got used to
standing in the @NordicTweets office, and I
miss that. Cabling on the floor is network stuff :/

Replying to @coderbyheart

And this project uses AD B2C authentication to protect a function app (which
needs to be configured against the AD B2C).

@Azure seems to not support CI of infrastructure
that involves setting up solutions from scratch.

Which worries me on how this project can be maintained.

Anyone knows a @SlackHQ bot that collects
achievements and notable items into a weekly report through emoji reactions?

I want to be able to mark certain things I see using an emoji and once a week
posts this as a document grouped by channel.

Reminder: ask your friends and colleagues you miss for a video call, they will
be happy to see you!

Also, feel free to do this during the work day, being at home means you no
longer have these random encounters at work.

Replying to @Maaikees

@Maaikees Same here, and this is incredible hard
to get under control. I have developed two mantras that help me:

  • I can't make people do what I want (even in a leadership role), I can only be
    the best example I would like to see.
  • I am responsible for how I feel, not others.
Replying to @ichaos1985

@ichaos1985 Nein, das war zwar nett aber auch
zu random, weil Leute zu unterschiedlichen Zeite dazu kamen. Das war immer etwas
anstrengended das gut zu managen. Ich mache lieber geplante 1 on 1 remote
coffees.

Replying to @fquednau

@fquednau
@NordicTweets We use both stacks (and other)
to build fronteds here, e.g. React on our Electron based nRF Connect for
Desktop. This role will be working more on our internal tools, which uses VueJS
(not exclusively, btw). We have a lot of freedom to pick the tech, so this is
not a must!

Replying to @tdpauw

@tdpauw I think many forget that it's not a binary
thing. If they have critical code, they can set up notifications for that, so
nothing slips through. But most code changes are fine, and speed has more
benefits than having 100% peer-reviewed code.

Replying to @ChristosMatskas

@ChristosMatskas
@Azure Yes, exactly. Basically all guides I see on
Azure features stop when it comes to infrastructure as code. (PowerShell and CLI
are not deterministic ways to set up infra). ARM is what I would love to see,
because tools like Terraform&Pulumi in the end compile ARM templates.

In normal circumstances I would be on my way to
@SoCraTes_Conf where I'd spend a long
weekend around many amazing people, meet friends and make new ones. Today I feel
💔 that we don't get this safe space where we can freely speak about our fears
and challenges. We could use that.

Replying to @piefke16

@piefke16 Roughly same problem as with HN: no
way to personalize the recommendations.

I want to bookmark a post, and if enough people "upvote it" (ideally for a
specific topic, and maybe with a friend algo), I get a notification.

Replying to @coderbyheart

I wish however there was a public statement on this process. We all make
mistakes, this is totally normal. By not acknowledging what went wrong the
organizers take away a valuable learning lesson from others who might do a
similar thing.

Having zero embedded development experience I am happy that the
@Espruino enables me to try out hardware project
ideas super quickly without any toolchain woes. It supports a small set of
@NordicTweets kits and I hope that one day
there is this experience for all our kits.
Embedded Photo

Replying to @coderbyheart

@m4nl5r @Espruino
@NordicTweets It's quite arguably that there
are many applications which involve external hardware that do not have a large
or actual unlimited power source. So trading low power with a device where I can
update the business logic on the fly allows for an entire different category of
products.