“OP uses a sledgehammer to build a bird house and complains it’s too difficult”
This is equivalent to using K8s to deploy a static site and complaining infra is the hardest of all. React isn’t the skeleton key for front end apps. Sometimes using simpler tools is the way to go and adding complexity when it becomes necessary
It still boggles my mind that, in an age where K8s is huge and K8s's reconciliation loop pattern is well recognized, Temporal has yet to put "reconciler" or "state-machine" in their documentation or descriptions.
I say this as someone who loves using Temporal and see a big future for them.
Temporal is a managed state-machine. If you're familiar with K8s, you can think of it as a dedicated reconciler. Plain and simple.
Do you want a resource to go from one state to another (i.e. user orders food -> food is delivered) regardless of transient errors and other hiccups? Temporal handles the plumbing of driving your resource to it's desired end state.
Do you have a workflow that has multiples steps? Can anyone of those steps fail for whatever reason and you'll need to figure out how to do retrying/cleaning up? Temporal handles that.
Essentially you can write your business logic without the worry of including the (often) boilerplate retry, cleanup, scheduling, etc logic. If you're familiar with K8s reconciliation and how it retries failed operations again and again until it reaches the goal, it's much like that.
Temporal can do much more than that, but that's its bread and butter and I'm trying to describe it simply so that people can understand. It is a bit of a paradigm shift and has a bit of a learning curve when you first start using it.
ctrl+f "state machine" on this page :) but yeah this is a very good perspective on temporal. we should talk, if you're interested in helping us tell our story better.
i'm actually interested in a longer form "Temporal vs K8s" comparison - can make a case for Temporal atop k8s, or Temporal replacing k8s (already had a couple customers talk to us about this). It's a nuanced story to tell and I'm not sure if it has a conclusion.
The company I work at runs a HA setup of Temporal on K8s, so Temporal atop k8s is a great recipe. I'd be curious to hear the reasoning for Temporal replacing K8s (or do you just mean replacing a piece of k8s?) Seeing as how as reconciliation is just a tool it uses for it's container orchestration, not sure Temporal would replace that but an interesting idea nonetheless.
yeah tbh this idea came from our users so i cant really speak to it but they really mean replacing k8s. this one for example is a large Jamstack/CI/CD provider with very spiky workloads. To them, k8s, ecs, fargate, etc are all too slow and declarative, they need to spin up + spin down (major motivation, for cost) infra way faster and do more custom stuff so they turn to us. i dont claim to fully understand it but does that give a hint?
Isn't this pretty par for the course when it comes to managed databases? Because the database is managed, they restrict super user access so you don't bork it.
Seems to be the case for AWS[1] and GCP[2]. Not sure if they have a workaround when it comes to `pg_restore` but my understanding was that if you wanted super user access to your database, you would need to manage it yourself.
Maybe CRDTs are useful for implementing chat systems since the operational semantics for it are simple enough to make a CRDT system for it?
The blog post I've read in the first link seems to say that CRDTs gets very complex when applying it to a domain where there are all kinds of different rich operations. I guess CRDTs aren't fundamentally the right solution for the collaborative document editing problem then?
I have implemented a CRDT system and have also blogged about it afterwards. So I can confirm this.
Joking aside though, CRDTs are still a pretty esoteric space and all the various blog posts came in handy when I was researching how to build my own CRDT.
Most of the resources I've seen on CRDTs are from whitepapers and those can be difficult to read if you don't have a math background. I gave a talk at a Papers We Love [1] explicitly because I found the academic papers a big turn off from many people interested in the space.
You can’t compare uni and bootcamps because they’re a false equivalency as someone else pointed out.
The sole purpose of a bootcamp is to get a job afterwards. That’s why they market their job placement rates so much.
Uni, video courses, and MOOCs are for education. Not everyone uses them to get a job, so job placement for a video course or uni compared to bootcamp doesn’t give you much info.
Anecdote: I did a self-paced online school to get my first software job. I did research on it in regards to job placement.
After getting my first software job, I have since taking many video courses and even went back to uni to get a degree in Computer Science. All of that was purely for further education, not getting a job.
I wouldn’t go to a bootcamp in my position because I don’t need it. They aren’t focused on education or academics but job training.
Wish someone told me before I got $300,000 into debt that my liberal arts degree wasn't going to help me get employment. My university was certainly not up front about that. And if they had been honest about the fact that the degree was worth fuckall in the job market I would have not wasted my money there.
I definitely agree, I do think uni degrees are severely over priced and don’t prepare people for jobs as much as people think. I have an oceanography degree that just collects dust in my parent’s attic.
But that in itself doesn’t mean what Lambda did was any better. If anything, they both need serious reforms.
Bankruptcy law should somehow apply to school loans after 7 years. Bankruptcy puts some of the onus of bad, ill-advised loans on the lender. Bankrupcy law partially aligns lender goals with borrower goals, as they then have a vested interest in your success.
The only people to who can afford to go to school to broaden their education without a clear job benefit are those born rich. In other words, the people who used to go to college pre 1970s.
> The only people to who can afford to go to school to broaden their education without a clear job benefit are those born rich.
I respectfully disagree.
I've spent an enormous amount of time inside universities (as both a student and employed as an engineer). The vast majority of students have no clear job benefit at the end of their degree.
Most students are studying something with no obvious job correlation. The largest schools at most second/third rung universities in the western world are humanities (which I'm not knocking, but the employment rate for these degrees into related roles is abysmal).
Honestly, it's really sad to see. I've spoken with countless third year humanities/law students that are completely lost and have no idea what to do as graduation approaches (about 20% of law graduates at my last uni went on to practice law). Oh, and they're crippled with debt (in both Australia and the US).
In my experience, the worse off the student, the more likely they were to study something with poor career outcomes (one of the worst offenders was the bachelor of business, which was a popular choice for students hoping to escape the lower/middle class but had atrocious outcomes). I chalked this up to fewer educated role models when they were growing up.
HN is fairly skewed towards tech. The tech-related courses (CS, EE, etc) have great employment prospects, but they're the outlier.
> Honestly, it's really sad to see. I've spoken with countless third year humanities/law students that are completely lost and have no idea what to do as graduation approaches (about 20% of law graduates at my last uni went on to practice law). Oh, and they're crippled with debt (in both Australia and the US).
> In my experience, the worse off the student, the more likely they were to study something with poor career outcomes (one of the worst offenders was the bachelor of business, which was a popular choice for students hoping to escape the lower/middle class but had atrocious outcomes). I chalked this up to fewer educated role models when they were growing up.
Grandparent said these people can't afford to go to school; I don't think what you've said contradicts that.
I think we completely agree. Unless you have a clear job benefit (like studying CS) you should only go to college unless you're already rich and it doesn't matter.
Respectfully, this is nonsense. Employers still very much prefer people with college degrees for many roles, even if the degree itself isn’t 100% relevant to the job.
A four year degree is better than nothing. But its much worse than a few years work experience in your career field.
Obviously there's a huge exception for technical college degrees like CS, science, etc which are worth a ton. This applies to degrees in business, liberal arts, etc.
Sure there are rich people that go to school for shits and giggles.
But “job benefit” isn’t exactly the same as “job placement”. I went back to school for improve my own CS fundamentals but not to immediately get a job. And many people do the same (go back to school for a promotion or for a raise).
And since those people exist, comparing job placement rates won’t tell you much since the data isn’t comparable.
Are you talking about just people in the CS program or all university programs? My first pass through college was in oceanography and I definitely didn't treat it as a bootcamp. I had no idea what my job prospects would be like after graduation. I just knew I liked the ocean.
In terms of CS, I'm sure most of them are going for a job. But the fact that those who aren't isn't 0%, it makes it difficult to compare.
Also, the reasons for people not graduating college and not graduating from a 3-6 month bootcamp will likely differ as well. 4 years is a long time. There are many reasons why someone might drop out. So again, comparing graduation rates and job placement rates between uni and bootcamp won't really tell you much useful data, especially if you're trying to compare which one prepares students better.
And just to be clear, I'm not saying Uni is better or defending it. I have my own with college [1]. But I see too many people trying to deflect criticism from bootcamps when those criticisms are very well justified.
> But the fact that those who aren't isn't 0%, it makes it difficult to compare.
It is irrelevant for the other 99%, so it doesn’t make it any harder to compare it.
> Also, the reasons for people not graduating college and not graduating from a 3-6 month bootcamp will likely differ as well. 4 years is a long time. >There are many reasons why someone might drop out. So again, comparing graduation rates and job placement rates between uni and bootcamp won't really tell you much useful data, especially if you're trying to compare which one prepares students better.
Sure, the comparison shouldn’t be that simplistic.
> “Danner, who invested $1 million shortly after Lambda School's Series A, compared the complaints to his own experience launching the charter school Rocketship Education, which received public backlash for teaching elementary-school students on laptops without instructors for part of their day.
"They said we were experimenting on the backs of children," Danner said.
"But when SpaceX launched their first five rockets and they blew up, was that OK?" he continued. "We're in a more high-stakes world of human development. Still, you can't say that you don't like the way things are but don't want people to try new things."
This is one of the problems of mixing VCs in the education space. “Move fast and break things” doesn’t work when the “things” are humans, not code and technology.
Also, last time I checked, SpaceX wasn’t promising 80% of its rockets would work back in its early days.
> last time I checked, SpaceX wasn’t promising 80% of its rockets would work back in its early days.
Is your criticism that these places are making false promises, or:
> “Move fast and break things” doesn’t work when the “things” are humans, not code and technology.
RCTs and placebos work exactly this way. People literally die to help us learn what does and does not work. We lost 8-9 months of COVID-19 deaths because of a system that needs to go slow, and yet had experimentation regimes that still put participants at risk.
I don't know of any verification method that doesn't require some degree of risk for the participants, but I'd love to hear one.
This is equivalent to using K8s to deploy a static site and complaining infra is the hardest of all. React isn’t the skeleton key for front end apps. Sometimes using simpler tools is the way to go and adding complexity when it becomes necessary