Hacker News new | past | comments | ask | show | jobs | submit | abrookewood's comments login

Have to agree - it's certainly refreshing to see someone acknowledging that their technology isn't going to work best in every case.

Can you tell me what RPG is? I assume a language that runs on mainframes??

Report Program Generator, an IBM language from 1959 designed as an alternative to using punch cards.

From a glance ... unlike COBOL which was invented in the same year, it does not seem to be widely hated - possibly it's even well-liked. But since it is a proprietary language exclusive to IBM it is quite unfamiliar outside their silo.


I'm wondering that as well. I don't get that suggestion at all.

I don't understand the point at all. If you need to add some condition later on, why not just add it then? What benefit is there to just marking out the spot where you might add the condition at some point in the future?

For their one here it's just the ability to rapidly comment/uncomment conditions in a query editor while exploring the data or debugging the query, and not having to worry about the leading AND or OR.

I've also seen it in code with iterative adds, for example:

  for crit in criteria:
      sql += " AND " + crit
No needed to add a sentinel or other logic to skip the first AND. I saw it a lot before people got used to " AND ".join(criteria).

That's right - it's just a quicker way of being able to comment/uncomment conditions when doing EDA or debugging.

Yeah. Especially when I'm trying to see what makes some query slow.

I personally don't use it too, but I think it's origins are not just readability, but from developing queries in a REPL like environment.

As you develop and are constantly creating / debugging queries where you often add new and or or clauses as a whole line, that becomes much faster to add and remove those same lines as they're a single shortcut away in nearly all text editors.


Yeah, so often do I have an EXPLAIN ANALYZE query.txt file I'm repeatedly editing in one window and piping into psql in another to try and make something faster. So I put WHERE true at the top.

It is a bit silly but I think it just helps with code readability some people.

Scratch is fantastic.

There are also a number of similar (block-based) tools that let you create your own custom blocks and see the code behind them - e.g. Blockly (https://developers.google.com/blockly)


"Drivers live or die by their rating" - really??? I have never given the drivers rating a moments thought. Do you seriously cancel rides because the driver doesn't have 5 stars??

If a driver's average rating falls below 4.6 Uber typically removes them from the platform. A rating of less than 5 stars is treated similarly to if a customer reported a driver for poor service.

Ahh, OK I had no idea that was a thing. I thought it was just for customer decisions.

I guarantee their algorithms factor in average number of riders who don’t rate anyone lol. Maybe they weight them. Maybe they grade everyone on a bell curve. Maybe they use it in C-suite meetings to determine strategies. I can just about guarantee they don’t ignore it, though.

They’re a tech company;a decent rating algorithm should be one of the most likely things for them to excel in, or they’d just lose to any company that does because all the good drivers they’re dropping are immediately signing up to work for a competing service.

tl;dr It is in ride-services’ best to retain good drivers, no?


In NYC at least, Uber has become the de facto way of getting a private driver to take you places.

Until that meaningfully changes I don’t think they care about anything but earnings.

Interestingly, that attitude is what allowed them to come in and replace yellow cabs as the de facto driver for hire option.


> Do you seriously cancel rides ..

It's also possible that the algorithm behind the scenes allocates drivers to demand based on ( proximity AND driver rating ).

If so then lower ranked drivers would be 'starved' out by higher ranked drivers.


Yes, rating is used for prioritization when proximity is equalish

Drivers with below 4 star ratings used to get dropped from Uber and Lyft. Not sure how a non-rating affects their score, but I'd assume the worst.

A non rating doesn't directly affect their score. But if a driver doesn't get plenty of five star reviews, they'll get dropped from the app because there are always people who give bad reviews because they had a bad day.

I've seen it plenty. A request to process an Excel file or generate a PDF etc. Basically anything generating or processing documents is a likely candidate. It might only affect a single application, but if you are running multiple apps on a box, it is often enough to cause an outage.


The biggest issue with this is when you deploy multiple applications to a server (e.g. 5 apps on IIS or whatever) and one of them kills off the box when it behaves badly. You can auto-scale, but it takes time to provision new machines and until they are up, you are down. Once you've experienced this a few times, the desire to split out applications into micro-services gets pretty strong in order to limit the blast radius.


No matter what auto-scaling solution you pick, it'll take time to start fresh new instances.


Agreed, but if you are running on micro-services and an app crashes, you don't lose everything (hopefully). It's not enough to make me want to use micro-services everywhere, but it is a consideration.

I'd like to see things work on Erlang/Elixir in Production and whether that works better, since the BEAM is very good at preventing individual processes from dominating the server and is also very good at recovering from errors.


Ah there's your problem. Don't use IIS on Windows Server 2008!


We don't always get a choice.


I'm sorry, what? IIS? o_O


The web server that many of us are stuck with unfortunately ;)


Had a very similar thing on the C64, but can't recall what the game was. In my case, the copy program would skip the bad sector, but I was able to use a sector editor to ready the bad sector with the original disk in place, then swap it out and write it back to the copy. Worked just fine. No idea why the sector editor was able to do that but the copy program wasn't.


I remember the feeling of accomplishment of getting the game to run. It was better than playing the game in fact.


That's strange, I had a friend that had a C64 and later an Amiga. I was always so jealous of how many more games he had access to than I did.


Amiga with Xcopy-pro is a friggin machine for copying some protected disks from other platforms (eg Atari ST)


This is the bit I found curious: "adding a small amount of random latency to requests to storage servers counter-intuitively reduced the average latency and the outliers due to the smoothing effect it has on the network".

Can anyone explain why?


Synchronized network traffic can cause incast or other buffer overflows.


Yeah jitter is generally used to mitigate “thundering herd” type problems because it reduces the peak load by spreading it out over time.


Thanks to both of you - makes sense


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: