dorianj

Stemma | Full-stack and Front-end engineers | Fully distributed (US and EU time zones) | Full-time | https://stemma.ai

Stemma is used by companies like Flexport, iRobot, Tempo, and many others, including public financial services organizations to help with discovering, understanding and trusting their data.

Data storage has been decentralized by cloud data warehouses like Snowflake or Redshift. Consumption by Looker or Tableau. Transformation by dbt and Airflow.

Organizations finally have the technical primitives they need to decentralize data consumption and transformation. But they run into a new brick wall: decentralizing understanding of what the data represents and how it actually maps to the real world.

Stemma helps solve that problem. We have strong initial traction but are still early stage: we're looking for curious engineers who are excited to rapidly iterate on new ideas with customers and build a product from the ground up. Every member of the Stemma team has the opportunity to meaningfully contribute to the direction of the product, business and the engineering culture.

We're funded by Sequoia. We're currently 12 people but are growing quickly.

I read every application sent to https://jobs.ashbyhq.com/stemma

Great question; I'm an engineer at Heap. This data model fit our needs at the time, however, you're absolutely right that these user rows do become quite cumbersome as they grow large.

> Well... generally this is solved via indices. An event log indexed by user ID would have the same effect. You get all the user info, and if you really need to grab historical or event data, it's there when necessary for parts of the app that actually need it.

Indeed! We're investigating a more traditional schema, as the topics of our more recent blog posts might suggest (http://blog.heapanalytics.com/speeding-up-postgresql-queries...).

While that's certainly true, the distribution of asteroids by size [1] suggests that increasing size by that much decreases the probability proportionally?

[1] http://www.aanda.org/articles/aa/full/2007/18/aa6046-06/img2...

Home prices are rising rapidly, but rent--even in central areas--is still quite reasonable for tenants.

Sure, why not.

Dry cast iron skillet, put in oven, heat to 500*F. Prep steak with olive oil and salt and pepper.

Take skillet out of oven, put on stove at high. Cook 30 seconds on each side. (total cook time on range: 1 minute)

Back in oven, 2 minutes each side. (total cook time in oven : 4 minutes)

Let rest 3 minutes.

What kind of FPGA package were you previously using?

Many FPGA CPU simulators attempt to be cycle-accurate for the purpose of architecture development, where the emulator has the leeway of treating instructions as atomic units and abstracting the rest.

Even more interesting ethically considering a large portion of Mozilla's revenue comes from Google.

I was thinking the opposite. I'm glad he's charging for upgrades, and charging a decent price for something some developers use all day everyday. I hope this will enable him to continue developing ST.

I'm growing fairly dependent on plugins, though -- I hope the new API isn't too hard for plugin devs to upgrade. Although with most plugins hosted on Github these days, compat is just a pull request away.

In the example URL you gave, the content of the URL (base64): ({:project-id "505a125e44ae42e05a750c97", :object-instance "2", :object-type "0", :device-id "1234"} {:project-id "505a125e44ae42e05a750c97", :object-instance "1", :object-type "0", :device-id "1234"} {:project-id "505a125e44ae42e05a750c97", :object-instance "0", :object-type "0", :device-id "1234"})

seems like it would be better stored on the server in redis or something (or, at least if leaving it in the URL, a more compact deduplicated format might be worthwhile)

The unlabeled referral link here at the end praising his host bothers me a bit -- shouting praises for a good service is fine, but at least note that you are getting a kick back.

> Because I know I will get an angry email or two, just let me say this. I know that Cloudflare touts other features such as security, load balancing, and keeping your site up if your server is offline. I did not test or take any of these features into account. Their biggest sector is to people that want to speed their website up, so I took them to the task on just that claim.

If this is true and people really think that CDN = faster site, then it's a misunderstanding, perhaps perpetuated by malicious marketing by the CDNs themselves.

But putting a proxy in front of an unloaded site (hosted on a relatively fast server) is of course unnecessary.

Re-run the test using `ab2' instead of simple 3-hit tests, and that's when the CDN becomes more useful. Or perhaps host the site on an oversold bargain shared server and see how it fares.

I don't have serious problem with this... but ballot stuffing and fake account registration seems the domain of SEO/blackhats or Anon. This isn't a 'cute hack' worthy of praise -- it's just abuse of MIT and Dropbox systems for no great reason other than perhaps pride.

PowerPC has a weakly ordered memory model, so older Mac programmers are no strangers to this sort of stuff.

There have been many multi-core ARM devices out in both the Android and iOS world for a few years, and I haven't heard too many complaints. I think most good high-level programmers know that they need to use high-level locks provided by the SDK, while anyone writing code that actually needs to deal with memory ordering probably already knows to use barriers some of the pitfalls of different ISAs.

This is needlessly complex. I really don't see pixel ratios of anything more than 2x being useful. Having two versions of assets is painful enough.

The apple convention, of having images with "@2x" appended, works quite well. The server could look for a http header indicating DPI and serve accordingly. This gives the developer to provide either one single HiDPI that's served to all clients, or separate files for HiDPI and normal resolution.

Yes. It's already somewhat common elsewhere in the world, by simply printing new barcodes for other sku's, sticking it on an expensive product, then hoping the cashier won't notice (they often don't in a store with a lot of different products, like Wal-Mart).

The charges are different, though, since it's fraud and not outright shoplifting.

And if the infringer doesn't reply to the 'friendly' request, the 'threatening' one can certainly still be sent, with the same effect as if it were sent initially.

I'd guess the display pixels are holographically projected into the field of vision at infinity?

Chrome is using the UIWebView control (same as Safari, sans NitroJS), so I don't know if that's even possible.

I've been wondering about this, if I'm the only one going crazy or what. I don't think the ones that appear in your news feed are called "sponsored stories", but I could be wrong.

It's almost enough that I want to email my family and tell them to unlike all the crap they liked... Even trying to get FB not to display "likes" from that person fails to hide them.

Specs say 216ppi? "7” 1280x800 HD display (216 ppi)"

The Retina iPad is 264ppi.

Not everyone drinks it for the caffeine alone. It's quite delicious when prepared properly. So, I don't think you're crazy if you're not crazy about coffee's flavor.

Yes, it is dangerous, and totally breaks basic firearm safety rules.

Most on-die embedded DRAM is one cycle away for a whole word (I'm not sure about arduino but I'll bet it's in the same ballpark). This uses a 500 ns pulse for each bit, so reading a 8-bit word would take 40 clock cycles (at 10 mHz)

Angry Birds is one of the most popular applications for the iOS platform ever. It has sales in the millions, and this model isn't comprable for indie software that doesn't have that big of a potential market.

Rovio also just released Angry Birds in Space, which is a separate app, meaning they'll get new income from it. But because it's a $0.99, giving existing customers a discount isn't important.

He mentioned he was in Texas, so this is indeed the case.

Thank you for being forthright and responsive.

I think the better thing to do for older devices would have been to make it a user-settable option, with a note on first install. I would have preferred to have a 5 or even 10 second delay in opening my 1password keychain, rather than have it less secure. But it's a moot point now--as others noted, the older devices are quickly falling out of favor (at least among those savvy enough to use something like 1password).

The crux of it: "1Password uses PBKDF2 to significantly slow down attackers. Currently this is not available on iOS as we needed to support older devices. The next major release of 1Password will only support iOS 5 and at that time we will be incorporating these additional defences"

This is absolutely unacceptable. Agilebits has long talked about how they use PBKDF2 to secure your passwords, and using anything different on mobile is an abuse of trust.

Agilebits' response is hand-waving. Of course longer passwords are more secure. If we all used 32-character passwords, we wouldn't even need key derivation.

But we expect to use reasonably simple (7-10) character passwords because it's possible to make these secure against cold attacks using math. And when the software you use has in the past described the algorithms it uses, you expect all versions of the software to work in the same manner.

I've long used 1password, and I don't plan to stop using it yet, but this is a serious breach of trust, and I feel less secure in using it now.

This is so stupid, I'm a bit surprised it came out of a normally excellent site.

The crux of the article is that retina graphics use more bandwidth, and thus Apple should have "thought about it" first.

By that logic, the first iPhone never would have existed, because it used way more bandwidth than any phones before it.

I've been using Sparrow with an app-specific password for months without problems.

I'm nearing completion of an EE at UT Austin, and this MIT class seems to condense 2 or 3 different semesters of material. I can't decide if this is a good approach or not.

On one hand, you take one broad class that introduces you to many different topics. Then you delve into whatever is necessary for your degree.

On the other, you do deeper (but still introductory-level) examinations of each topic, still with the option to go deeper if you desire.

Ultimately, I don't think either is a clear winner. I don't remember enough out of my intro circuits class to do a thing with a second-order circuit without reviewing theory for a bit first.