Any revision of your thoughts on Angular v React now you've done a lot of React?

From what I understand Angular 2 nicked a lot of the best ideas from React. I have an Angular 2 pet project to do that I still haven't got around to.

+1 on being very interested in candybar's evolving react vs. angular feelings as he learns more and more react.

I am doing a React POC now just to get up to speed. Found these videos which seem pretty good: https://www.youtube.com/watch?v=MhkGQAoc7bc I love their ES6 videos - they really hit home how most of the new features make your life much easier.

Any other recommendations for learning react?

Btw I did some polymer POCs based on these tutorials: https://www.youtube.com/watch?v=MaWcS-10NIw It seemed pretty cool but I get the sense it's kind of dead now? I love "Scott with LevelUpTuts"! He's so peppy.

I enjoyed the polymer stuff, and still think web components are the future, and angular 2 is a lot closer to that than react, right?

Story time: I just checked in with an Australian developer I worked with on my first web programming job in 2000 - a psycho startup in LA. This guy was by far their most prolific dev. He had written like 2/3 of the entire codebase, out of 10 devs by the time I got there. I think he was 24 at the time when I was 31 or something. I basically learned java by pestering him all day.

So I checked in because the startup I'm working for are all in Aussie and I thought he went back at some point. I figured he'd have to have learned new stuff. Here's his reply about what he's doing:

Dude is way too young to be a 'legacy developer'. He's younger than me and a better dev than me (or at least he was at the time and probably until I got 10+ years of experience under my belt). He should be building amazing things.

Moral: stay current kids, and move if you're not being pushed into new stuff. I've gotten really really lucky in that wherever I went there was always some cutting edge thing going on that I got pulled into.

I read somewhere that the W3C committee writing the Web Components specs is stacked with Google guys. Angular 2 was intended to be a transition to web components and Google are writing both it and the Web Components specs, so yeah.

Yeah that was my impression. I still don't like JSX very much. But I'm trying to give react an honest trial.

They're hoping to use react native to save a lot of effort on the iOS app version of our site (mobile web is not an option). I've heard some bad things about react native though.

So far I have Angular 2 > React/Redux > Angular 1 but admittedly I don't have that much experience with anything other than Angular 1 out of those 4 technologies.

This is something I've believed in and speculated about but I'm becoming far more certain now that I work for a reasonably large tech company - platform choices and architectural decisions for a large company at this level are much more about people and organization building, whether in terms of recruiting and retention or simply corporate structure (formal and informal communication structure) that these kinds of macro decisions influence. Efficiency of the sort engineers think about simply doesn't matter much because at the macro level, if you are able to recruit brilliant people, keep them motivated and have them communicate efficiently amongst one another, they will figure out a way to make things work even if lots of things are done poorly from a technical perspective. And this can be very contradictory because technologies that help you solve problems you have today most efficiently can nevertheless make it difficult to hire some of tbe best people and/or force you to structure your company in a way that leaves you less able to tackle future problems.

Check the query. If it is optimal, great.

Check if the index is firing. If so, then that may well be the fastest that query can run.

If the index is not firing, then there are memory settings, probably something called "buffer size," or perhaps a memory leak: http://stackoverflow.com/questions/2...so-much-memory

Indexes aren't the end-all be-all of any system. They are used for certain queries, but not others. Indexes are add-ons, not crutches.

Finally, Stack Overflow.

Welcome to the real world bro, unlike your one-man show where you got to optimize everything to infinity. When we started putting node and angular on our job requirements, our candidate quality went up a ton.

Thanks dude, I passed this along to him as well. They all swear this guy is really good but I'm starting to wonder. It's never good when you get contradictory garbled response. IE after I sent him your first advice I got this response:

To which I replied:

To which he replied (he's French btw, in France):

Ok so you didn't actually solve the problem?

Btw the github ticket is very convoluted. They tell him that his problem really isn't the same as the thread and he says he'll submit another ticket, but he's still quoting to us that he is waiting for them to reply. :/

Yeah, tbh when I talk about "better" and "worse" I am mostly talking about what I would most like to solve problems in. My question amounts to "Will Angular 2 or React make my life easier?".

@suzzer; I know you know I'm beating around the bush and I'm trying to be not overly subtle on what I think the real problem is (bad schema / dirty dataset). Do you think the conversation and the github post is pointing in that direction?

I was looking through some of the source files and ran into this:

https://github.com/rethinkdb/rethink...lient_protocol

" Raising query limit to 128M for #4529 "

issue 4529 specifically discusses binary and table sizes: https://github.com/rethinkdb/rethinkdb/issues/4529

https://github.com/rethinkdb/rethink...ment-120578254

Thanks for the report. It looks like you're hitting the query size limit. r.binary was designed to hold smallish chunks of binary data, not large files, so in general the way to solve this is to split your files into chunks.

Doesn't quite make total sense because your developer is claiming his chunks are 1Mb, but I'm wondering...

I'm quite impressed that a maintainer is willing to go so far out of his way like that. With so much concern on what users are doing that and being backed by YC, it may end up with a pretty decent future.

I'm definitely interested in how this resolves itself.

Yeah, exactly, I have been insulated from certain realities because I was at a small company for too long. And I think this goes a lot further - in some sense running an elite software engineering organization at scale and general principles of engineering good software clash dramatically. Think about the kinds of things we kind of take for granted as good principles for architecturing software.

Law of Demeter
Single responsibility principle
Open/closed principle
Liskov substitution principle
Interface segregation principle
Dependency inversion principle
Modular design
Explicit State/Change Management
Non-leaky Abstractions
Separation of Concerns

I'll call these qualities "modularity" for short because I think many of them are about the same thing. Essentially, we want software components to be predictable, have local impact, easily isolated, replaced. Given Conway's Law, you can map, to a degree, software architecture to organizational communication structure. And one thing becomes really clear - talented engineers absolutely hate and detest organizational modularity. They don't want to be limited in terms of responsibilities, they don't want to know just what they need to know, they want to be able to touch everything, they operate simulatenously at multiple levels of abstraction and have a holistic, as opposed to a local view of the system.

I personally haven't really tried to evaluate frameworks from this perspective before - how will adopting the architecture prescribed by this framework shape our organization and prepare us for the future - and should have more thoughts on this later.

I have been working on my Mac Book Pro (2560x1600) with no extra monitors the last ~2 years. Always thought about getting an extra monitor but never did. I used to rock a tripple monitor setup when I still played poker...not really missing anything.

I'd probably be too tilted if I had a multi monitor setup and then moved around to work elsewhere. So I standardize on that one monitor. At home I like to work in the garden, too...would be inconvenient.

I'd be interested in other opinions since I think about just buying one-two good extra monitors about once every 6 month but ultimately decide against it.

Interestingly I'm mostly familiar with this through the business part of my business/CS degree. Maybe reading a project management POV helps? Terms to google: precedence diagram method, PERT, CPM, GERT
https://en.wikipedia.org/wiki/Precedence_diagram_method
https://en.wikipedia.org/wiki/Progra...view_technique
https://en.wikipedia.org/wiki/Critical_path_method
https://en.wikipedia.org/wiki/Graphi...view_Technique

Unfortunately the Englisch articles are a lot worse than the German ones...the first one by comparison:
https://de.wikipedia.org/wiki/Netzplantechnik

It was actually fairly interesting to do these on paper for exams. This and operations research were the most interesting business classes (I also liked strategic management).

If you finish that exercise, there'd probably be some interested in a decent open source tool that spits out a diagram from some spreadsheet entries. I could use it but I have always been too lazy and just used lame bar charts for project proposals...I think a precedence diagram is a lot more useful as a planning tool.

Yeah I'd say this guy is a keeper if that's his first piece of JS code and it's not copy/pasted from somewhere. I think complaining about that piece of code from an outsourced dev is pretty high level complaining...then again we always did inhouse and I think it's a horrible idea to treat programming as a cost center.

Also there was a reasonable argument for using underscore map/filter etc. over the stock versions but I forgot it...it's in the functional JS book...I'll look it up sometime next week.

Does anyone have an opinion about using SQLAlchemy in python?

a) I know SQL extremely well. I'd say I'm an expert.
b) I don't think I'm going to use it for an ORM.
c) Looking at the docs, the query syntax looks pretty torturous compared to just writing freakin sql queries

It has a very good rep. What's the benefit? Being able to use it with multiple databases? (not really a problem for my use case)

What do the guys who have looked at angular2 think of the rxjs stuff? I've tried to figure out what the .publishReplay(1).refCount() stuff was doing and every time I fail to get it. In general it just seems hard to understand for the simplest things (a stream of functions that you subscribe to and then apply to some data in another stream to maintain state). I can't imagine how incomprehensible it could become for anything significantly complex... but maybe once it clicks it's actually easy?

Did you find the textual syntax? It doesn't look significantly different than pyscopg2, and perhaps a bit cleaner:

http://docs.sqlalchemy.org/en/latest...ng-textual-sql

psycopg2 looks like this:

serious question, what is the point of code review if this is the kind of stuff you would pass?

Yeah I get that you can just run raw sql queries and that seems fine, but then I might a well use psycopg. I guess my question is, is their fancy query building worth a **** or not (to me it does not seem more readable than straight SQL, although it does at least paper over some of the inconsistencies that are kind of annoying, especially when you are building a query by pieces)

I designed our node framework to isolate the business logic, which is the only thing any new/inexperienced devs should be touching anyway. Obviously we still code review, but at some point you're not on the project anymore or it gets too big or something slips by. At least they're only writing crappy code in their little sandbox. I still keep a close eye on the framework code and only one or two devs ever touch it.

I think for big companies with giant sprawling websites who insist on using offshore devs and therefore have a lot of developer churn and varying skill levels - this is really the only way to go. YMMV.

Although knowing what I know now, I will always try to design something like this, even for a small team full of rock stars. It just makes for such cleaner, easier to maintain code. It's one of those things like testing, that's a little more work up front, but once you come to rely on it you can't imagine not doing it.

I've been told it's not as easy to isolate business logic from "plumbing stuff" in the angular front end code. But I also know our angular lead is really into just getting stuff to work, not designing scalable frameworks to keep junior devs out of trouble.

So I'm not sure if I was the "angular guy" as well as the "node guy", if I could come up with some kind of business logic sandbox for them that abstracts out most of the repeatable/low-level stuff.

i can't speak to its use in angular2 (though angular seems to have a knack for over-engineering), but conceptually, that Rx stuff is great and the paradigm will indeed become simple to understand once it clicks.

have a look at elm, cycle.js, and probably start with this The introduction to Reactive Programming you've been missing

I've never used SQL Alchemy, though I've used other DSLs. I didn't like the generated SQL. For many cases, they use left or outer joins instead of inner joins, passing much of the work to the query planner. I'm concerned that the example I posted has some bad practices, so I'm guessing that's the target market?

in regards to building up queries, the processing language has some interesting ways to construct queries via format('select * from %I', my_table), which works for columns as well, though I don't use that stuff very often. The optimizer can't see inside of a function.

In short, see what the generated SQL looks like. If it's awesome, I may decide to start using SQL Alchemy, as long as it can handle windowing and CTEs.

It is supposed to be able to handle windowing, I have not tried it. No idea about CTEs.

What I actually can't figure out is if I would even prefer building queries using their tools or not. Like, let's just assume it creates perfect SQL. Would I want to use it even then? I'm not sure.

I mean, like...

Did I just learn a language that is nearly parallel to SQL just so that I could do something that's already pretty easy to do directly in SQL?

Why? That's horrible to look at. If someone starts with syntax like that, then once they are forced to use some other driver, they are not only writing unreadable and horrid SQL, but worse, prone to making terrible security mistakes.

I wish python had an analog to HugSQL for clojure: http://www.hugsql.org/#using-ddl

I'm not entirely sold on the idea of using comments to define function names, but it is clean to look at and simple to understand.