haggy

First, congrats on getting healthy!

Secondly, since you're presenting this as a portfolio entry, I have some open, honest feedback. I quickly reviewed the repository and there are some pretty major orange-red flags that would dissuade employers from reaching out. Some of them include:

- Lack of coherent repository structure. All files under src/ with no sense of modularity

- The commit messages are very poor. Messages like "cool" and "jump jump" and "perf" do not provide any context to anyone outside of yourself (and probably not even you will remember what those changes were days/weeks from now)

- There are magic numbers everywhere. The biggest offender I found was in shaders.ts

There are more but those are the top ones I saw from my quick review. Again this isn't meant to discourage you. I wanted to provide the feedback to help you be more successful with potential employers. Cheers!

Kafka, unlike Mongo DB, relies on recovery/replication instead of fsync:

https://jack-vanlightly.com/blog/2023/4/24/why-apache-kafka-...

Kafka has never tried to hide that fact and it does not, in any way, make Kafka unsafe.

This ^^^ My answer to this question would have been a VERY weak "Yes" before the Twitter disaster. Now that we're seeing the true Elon alongside real engineer feedback, it's a HARD no.

That totally makes sense. I think you'd have a miserable time supporting this at scale if a re-compile is needed every time a new workload requires a tweaked setting or two. Config based static allocation is the best of both worlds.

I disagree in this case. Normally I'd fully agree with you but I think this animation style works really well given their use-case.

This is not really surprising considering that it's basically a (very cool) marketing article for their equipment.

They explain pretty clearly that this is static memory allocation. In other words, a calculated amount of (mostly) contiguous memory is allocated at startup and used very carefully, relying on disk to supply the dynamic nature of database load semantics.

https://www.geeksforgeeks.org/difference-between-static-allo...

Per the article: "(Currently it’s determined even before startup, at compile-time, but startup is the important point since that is the first time that memory can be allocated.)"

Sooo are we sure you can even change this via proc restart? This is saying the limits are built-in during compile...

Not a fan of what feels like an advertisement hijacking the OPs post. Most would probably prefer you just make a separate HN post.

Ah thank you. I figured the 'Q' stood for quantum but you saved me a fair amount of googling :)

The magic of PR's :)

The article body and headline are (of course) at odds IMO. The headline is very much "doomscroll" material while the article points out that interference could basically manifest as having to rely on secondary systems and protocols to land. If I'm missing something then happy to be corrected :D

Depends how you group them. If we're talking battle tested then yea maybe four but I believe there are many more that don't have widespread adoption

> A high rating is not enough to guarantee a high auction price - although it helps. Another copy of Super Mario 64 in the same auction with a 9.6 A++ rating sold for a comparatively affordable $13,200

This seems to reinforce some kind of laundering/tax evasion scenario. There is little logical reason for one cartridge to go for $13K and another to go for $1.5mm when both are certified excellent condition. It appears that the one going for far more is a product of timing. Someone needed to move a large chunk of money quickly and decided on the Mario cartridge.

Can you imagine the CI build time bloat? I'm getting anxiety just thinking about it

I don't know if I'm following. Are you saying you don't believe this change is a big deal for the climate?

How do you figure? Http proxies support various common cache invalidation strategies such as hash-TTL and ETags

That could be interesting. Update us on how it goes :D

No offense but just "running a neural net" is rarely trivial

What useful software exists which works on no data at all? Even a calculator requires data

Moderna is well known to have more injection site pain and greater immune response for the second dose but it's showing vastly greater efficacy percentages than all other manufacturers

That's why most programs like this send both a social worker and police officer. The social worker initially takes the lead but if things escalate past the point that the social worker is able to de-escalate the police are there to intervene.

Yea they even mention that traditional policing didn't go away, they are just now focused on crime

What issues did you run into on the JVM with Java/Scala? I spent 5 years working up and down the entire akka stack so I'm curious to learn from your point of view. Also did you try akka typed or classic?

Local NAS with offsite backup is far more resilient than any physical item onsite. One house fire and that album goes up in smoke.

Interesting! I'm going to read into the info you linked. Thanks for the info!

Can you point me at documentation for the fault tolerance of the system? A huge issue for streaming systems (and largely unsolved AFAIK) is being able to guarantee that counts aren't duplicated when things fail. How does Materialize handle the relevant failure scenarios in order to prevent inaccurate counts/sums/etc?

Oh wow I worked on a project over 5 years ago that insisted on RethinkDB. Man so many issues... Lol

I don't know your background but the way you respond makes me think you have not been responsible for systems that multiple tenants rely on for varying workloads.

These assumptions you're making are dangerous because the variety of workloads across tenants is extreme. If you're going to do something like "kill compute no matter what" then you better have a good reason for it.

Wow what a read. Thanks for taking the time to write this.