rafekett

I don't think it makes sense to choose your logging framework based on performance. Logging does IO, so it's relatively slow compared to everything else you do -- but operationally it's invaluable. I'd take the performance overhead any day.

If you have to ask how much logging costs, you probably can't afford it.

A friend was doing this in Philadelphia last year and has started doing it in SF http://tea.byankit.com/. Not sure if this idea was inspired by that or an independent discovery -- the design of the pages looks samey but I think basically all designs that have the same color palate from the same year look samey. It's just whatever's in that year.

Perhaps the most interesting and useful fact about the harmonic series is that the nth harmonic number (the partial sum of the first n terms) is an extremely accurately approximation of ln(n) (they are very close for all values of n, in the limit they differ only by the Euler-Mascheroni constant, which is about 0.577). This is the only approximation I have seen that has constant error in the limit.

I did a similar analysis a few years back. http://rafekettler.com/2011/06/16/analyzing-the-lulzsec-pass...

I'm sure his project worked, too, it just didn't scale.

On the plus side, this makes it very pleasant to travel to Argentina if you're from the US. Just don't forget cash.

Starbucks is involved with a surprising number of big SV players.

All institutional investing today is based off arbitrage. The only thing strange about HFT is that a computer does it.

> That they are not allowing this for credit cards indicates that they are using the pin debit networks.

not necessarily. you can't think of any other reasons why they'd do this?

I worked on a similar system at somewhat larger scale (maybe 5-10x larger volume and dataset size). There's a lot of basic JVM tuning to be done. You have to monitor heap usage and GC pauses and keep manipulating heap size, perm gen size, heap usage that should initiate major GC (forget what variable this is) until you get things as smooth as possible and meet whatever resource constraints you want to meet on your hardware. In my experience, for large datasets ES really pushes the memory architecture of the JVM -- it doesn't really perform too well when your heap size is like 30g.

After that, you'll have to tune merging. The underlying Lucene storage engine chokes when it tries to merge segments that are large, so you have to tune the max segment sizes, etc.

Then, you'll have to tune your queries -- it's not feasible to do a full index scan on a large index so you have to get clever about how you pull data in chunks. If your data is timestamped and that timestamp is indexed, you can pull data for smaller time ranges which will be faster than pulling all data for an index at once.

Leader election in addition to what else has been said

Have you used ML?

the choice of projects seems somewhat arbitrary. i've only contributed to one (python), but i can definitely say that many of these codebases are total messes (drupal, libreoffice for sure) which would make me _not_ want to work on these projects. in reality, the easiest way to get introduced to open source is to start your own projects to build the kind of reputation and experience necessary to contribute to bigger projects.

The idea of a minimal technical hurdle is important in free software. If we adapt everything from our free world to look just like the Windows world, Windows users will never notice that there's something else better. It's important that free software give itself some competitive advantage over nonfree software, especially for its users.

I don't work on anything with a README that is for nontechnical users, and I don't produce technical material for anything other than UNIX. No one I work with that I care about pleasing uses it. I'd rather stick with the tried and true convention.

I actually did build all of my company's push stuff for iOS in an afternoon. It's a very simple TCP protocol that's well documented. If you know anything about basic networking it's cake.

Sorry, but $6500 for 1/2 of your company seems like a raw deal at any stage. If your idea is even halfway decent, you should be able to do better than $13000 for your initial valuation.

Protip: find real investors who are interested in giving you a fair price. This offer is basically equivalent to a payday advance or a loan shark in terms of cost.

This is a great piece of software in terms of presentation and ease of use. However, it can make your life difficulty if you want to do anything outside of the scope that the authors intended. It's very difficult to build dynamic UIs with chosen select boxes because there's zero documentation of the code itself, which is what you'll be modifying if you want to do anything advanced. The devs did a good job building a select box, but a bad job making it extensible and documenting it.

Overall, I'd recommend it if your needs are simple, but you might be better off rolling your own or using another library if you need anything advanced.

I agree. What I wrote here was based on that.

I wanted my post to be more akin to a tutorial than a reference. I don't think the data model reference is useful to people who don't know the basics -- the idea was to bridge that gap so that it could be useful.

And thank you for posting. I'll take all of the traffic I can get :)

This was my work (not sure why someone randomly posted it here, but such is the internet).

I do exactly what you just described, see http://www.rafekettler.com/magicmethods.html#appendix