HNHacker News
TopNewBestAskShowJobs

nhoughto

451 karma · joined October 9, 2014

submissionscomments
nhoughto··on I’ve been reading a whole lot about tomatoes
and for something slightly more accessible and with photos =) on the same topic:

https://www.seriouseats.com/why-you-should-refrigerate-tomat...

nhoughto··on Comparing ClickHouse to PostgreSQL and TimescaleDB for time-series data
Does timescale do it’s own compression alg too? I see in pg 14 toast column compression can be lz4 instead of ootb pglz which has a few probs appr, I see mentions on the mailing list of significant possible optimizations. When dealing with EBS style storage where read latencies can be multi millis compression is always going to be a win, but is an easy optimization either way I’d think.
nhoughto··on Comparing ClickHouse to PostgreSQL and TimescaleDB for time-series data
Ah so only costs 1 row for pointer and 1 row for toast? Well that’s much more deterministic
nhoughto··on Comparing ClickHouse to PostgreSQL and TimescaleDB for time-series data
yeah very interesting, i was wondering how timescale pushed postgres more towards columnar without rewriting a bunch of postgres itself.

My understanding of TOAST is that it itself is just a bunch of rows in a toast table that split the compressed "row" or in this case "1000 rows of 1 column" across as many rows as required to store the data whilst remaining within the postgres page size limits (normally 8kb).

With the often quoted postgres per row overhead of 23 bytes~ which you would have to pay for each TOAST row as well, does this not add up and eat into your storage efficiencies? or does compression work so well that the 23 bytes x N rows (1 row pointing to toast + N toast rows) required to store the "row" isn't important?

nhoughto··on Launch HN: Rainforest QA (YC S12) – No-Code UI Test Automation
Agree, no-code seems to optimize for a large number of people being somewhat effective, but with less control over the output (increasing fragility, rework)

Code is more upfront effort but more control and thus less fragility but more maintainable over time (if done right).

I imagine there are some/many situations where throwing many people at a problem is the “best” way and this would suit that quite well I guess.

nhoughto··on Single sign-on: What we learned during our identity alpha
suuch a hard problem, esp for government that has to include 'everyone' and all the weird edgecases that forces on you.

Couple that with proof of identity being one of the few things that might have been issued 40, 50.. 60+ years ago (birth certificates) and never updated (unlike passports..) and when issued had no concept or sympathy that it might be used for digital verification in the future.

Without redefining the problem to avoid issues like birth certificates, i'm not sure this is solvable in the way stakeholders expect it to be. Stakeholders (like citizens paying for it) expect to be able to point technology at the problem and have it solved, 'its just an app right?' or 'lets use biometrics!' but so many other things have to adapt to make something like this successful.

Estonia had an interesting approach where they just issued everyone smartcards/certificates and used that as proof, this bypasses the 'birth certificate' problem but is expensive (Estonia has smallish population, newish government so ok) but such an approach itself has a root of trust problem.. who do I issue the smartcards to? It isn't directly transferable to other countries/governments.

You work around the root of trust problem by creating exceptions and alternate paths to verification, maybe do it in-person for people with disabilities etc. But then how long does that take to roll-out? How easily abused is such a system? How many of the identity moment can I use it for? Is in-person verification trusted less?

It is very easy for the resulting system to be quite brittle too and not reflect real use-cases, real world identity moments are very diverse and often have more flex than you'd think and it is very difficult to carve out the right chunk of the identity problem to solve and which to leave behind and still create something that improves the ecosystem.

learnt through pain, built digitalid.com

nhoughto··on Gitlab from YC to IPO
yeah there seems like there is some web -> idea connection that should exist for super slick in-IDE reviews. Best reviews means most context, IDE has the most context about the code (is why its best to resolve merge conflicts too) so we should be leveraging it more.

Always surprised to see people resolving merge conflicts anywhere other than their IDE, IDE has the most knowledge about the inputs to the merge and your chosen resolution, why would you ignore that important info? (like.. does the resulting file compile? for ex..)

nhoughto··on Rancher Desktop, a Docker Desktop Replacement
Good to see people stepping in after the docker desktop change, things I care about:

- transparent networking, i want to access kube services on 127.0.0.1, i'd rather not run a proxy/manage things myself - access to containerd socket to use non-rancher tooling, for image building etc.

early days for this project, i'll have to spin it up and see how it goes!

nhoughto··on Cloudflare R2 storage: Rapid and reliable object storage, minus the egress fees
I've always found pre-signed URLs fragile and a pain to work with, when we solved this problem we just put cloudflare workers in front of b2, so the worker can sign requests to b2 itself, so the interface to the user can be simpler. Could just be a straight POST with bytes, then the worker will turn that into a PutObject call to the s3 api, it works pretty damn well.

If you are reading files back out again too you can use cloudflare caching to speed things up, its a good combo.

nhoughto··on AWS SIGv4 and SIGv4A – How AWS signs and verifies API requests
I was pretty sure that you can approximate this in aws iam just with policy conditions on “aws:CurrentTime”

But after much googling it appears not! Seems like a fairly simple additional variables aws could pass through and condition to support, so you can say “only allow this request if the signing time is < 1min ago”, closest you can get is that the policy is only valid after X date! Is a shame!

nhoughto··on ClickHouse, Inc.
nope just an interesting test we ran, thought i'd share. we actually did all our testing and then came across firebolt very late, so thought others might now have seen it either.
nhoughto··on ClickHouse, Inc.
We just did months of testing on a bunch of dbs for a time-series workload, and whilst we really liked the story and devs behind clickhouse. The ops burden of not separating storage and compute ended up being a turn off. Good to see it is progressing and will hopefully get more investment, although I wonder what this means for companies like Altinity.

We compared Timescale, Clickhouse, Snowflake and Firebolt. Ended up really liking Firebolt, some amazing tech with a few roughedges (its pretty new), basically Clickhouse speed meets Snowflake simplicity definitely one to watch.

https://www.firebolt.io/

nhoughto··on The never-ending product requirements of user authorization
Related and very interesting: https://news.ycombinator.com/item?id=28543457

Good to see more things happening in this space.

nhoughto··on Adopting the OpenAPI schema to generate Plaid’s SDKs
Yeah takes a lot to make a generated sdk good, lots of bad examples, few good ones.

AWS SDKs must be generated and they are mostly good but some of them like S3 are so heavily used they could have teams writing them no probs. GCPs are (or were?) super verbose which is a GRPC generation artifact, same as openapi.

nhoughto··on VisiCalc: Information from Its Creators, Dan Bricklin and Bob Frankston
Is there any good books telling this story? Seems like a lot of the detail is around but a consolidated kindle read would be great..
nhoughto··on A monorepo misconception – atomic cross-project commits
Don’t agree with the conclusion or reasoning, sure you might need to take a multi Stage approach for many reasons. But if you’ve got your shit together atomic change is possible, in many other approaches it’s not even nearly possible. Whether your org can put together the maturity / investment / whatever to make it a reality is upto you. But the possibility is the sell, not the guarantee. If you can make it work (and google do on a crazy scale) it’s an incredible power, incredible, like change your entire approach to tech debt incredible. Also has costs at scale like the theory about how google keeps shutting down products because keeping them in tree is expensive.
nhoughto··on If you want to transform IT, start with finance
Seems very familiar.

Its not that finance get things done, its that they can prevent things being done entirely, or reduce funding/investment to the point they naturally peter out.

nhoughto··on The James Webb Space Telescope has passed the final mission analysis review
And you’d need to have it designed to be serviceable once deployed. I’d imagine there would be a few assumptions going into the design process, “once deployed humans won’t ever need to get to that..”
nhoughto··on Product validation frameworks are mostly useless without taste
This summarizes some problems I see in the day to day with ex Amazon folk trying to transplant ideas without understanding why the process was successful in the first place. Some good detail that might help with that.
nhoughto··on GitHub Copilot
I wonder if there should be a thumbs-up / thumbs-down mechanism to help teach it, or even flag snippets as bad/insecure/whatever.

I'm picturing the StackOverflow problem where the accepted answer is actually wrong, and actively pushing it into more peoples code is just proliferating the problem into more places. Making people faster at building mostly the write but sometimes the subtly wrong thing.

nhoughto··on Bflat: C# as you know it but with Go-like tooling
Cool, so like Graal for Java. 2-3mb min file size is quite a bit smaller than the smallest Graal native images tho, I wonder what the difference is.
nhoughto··on WD My Book users wake up to find their data deleted
Yeah difference in threat model of “evil internet can make tcp connection to me” vs “basically need a c2 compromise” is huge. Sucks for those that lost data either way.
nhoughto··on WD My Book users wake up to find their data deleted
doesn't appear to be a directly connected to the internet problem, even NAT'd stuff wiped data.
nhoughto··on My EC2 Wishlist
Yep thought the same thing, reminds me of kube2iam, IRSA is a much better solution and obviously where AWS want to go. So yeah this one isn't going to get changed.
nhoughto··on Highlights of new technologies introduced at WWDC 2021
Biggest problem with iOS in CI is limited build primitives, some providers have macOS options, macstadium is a thing, Anka helps but it’s all so hard! Waaay harder than building stuff on Linux. Apple can wave it away but making it harder means iOS CI is going to be behind forever.

They had a chance to try and help close that gap and even charge for it, I’d rather not have to setup macstadium, configure Anka controller and pull 20gb+ images.. just solve my problem for me..

nhoughto··on How a Jenkins Job Broke our Jenkins UI
Always surprised when I see people still using jenkins, must be because of history? you wouldn't choose it today..?
nhoughto··on Wyoming site of new nuclear power plant from Bill Gates' TerraPower
Am I wrong to think this is progress? Always hard to know how likely these things are to actually work out.

I was disappointed that TerraPower in China never got going, bc trade-war. The inside bill's brain episode about it all was great.

nhoughto··on Sublime Text 4
I still need to upgrade from version 2, have been ignoring the update now prompts for a while now it seems!
nhoughto··on Extreme HTTP Performance Tuning
I’d love to have the time (and ability!) to do this level of digging. Amazing write up to, very well presented.
nhoughto··on Vcluster – Create fully functional virtual Kubernetes clusters
the only problem with this is needing to push your container images in CI before being able to schedule, as the real pod will a real node, which will need to pull a container.

In kind you can build container images into the single host, and avoid a container push/pull cycle. Any solns to that in vcluster?

← PreviousPage 6 of 9Next →