Shoe Lace - what's your answer for "what does new do"?

It creates a new instance of the object you pass to it. The instance then inherits from the object's prototype that you passed in.

--

On a side note I'm pro using new and I also like to setup my objects using prototypical inheritance. I found it makes the most sense to my brain because I internally map it to classes and methods.

So after fiddling with my network configuration for a couple of hours and still being convinced that everything was configured correctly I decided to say F-IT let's give DHCP a shot once before monitoring everything with wireshark (it should be noted that I'd have to read up on this first).

Boom like magic stuff just works...somewhat unsatisfying but I guess better than nothing. What does DHCP provide beyond just assigning IP addresses that I might be missing?

[As an aside I also learned a bit about programmable switches and vlans so that's good I suppose]

Either way I have now finished all the technical stuff I wanted to get done before the year ends. My ownCloud now works via SSL on my webspace (+WP is set up with SSL as well so I can start blogging) after finally getting the cert...dropbox already dropped. FU NSA

And I have a semi-interesting idea for my Pi for the next year...hook it up to a nice monitor/keyboard (bonus points...touchscreen instead) and run Trello in kiosk mode. Heyachachacha techy Kanban-board.

[whine]

Setting up a mail server is biggest pain the ****ing ass ever.

Sure, it works sort of, until the whole server crashes because there is one thing not pointing to the correct place. Oh, of course, mail@mySite resolves, but it doesn't resolve from the server, and then all tests pass, but it still doesn't work. And then make a change, wait oh, idk, 5 hours for all the changes to propagate, and then wonder if it it isn't working because I really screwed up or the changes didn't propagate.

[/whine]

Dev Ops, SysAdmin, whatever them folks are called that uses those silly Happy Hacking keyboards are worth every last penny they juice from their company. Takes a certain type.

Dave, are you setting up a relay or the whole thing?

I heard a lot of horror stories about mail not being received because you're not white listed on certain big name providers. Ridiculous amounts of spam too apparently.

I've spent the last couple of days really getting into "dev ops". I'm not touching mail servers but I'm really digging automated deployment and server management tools.

I feel right at home with chef/berkshelf. Only problem is my computer isn't well equipped with running more than 1 virtual machine (so I can test the deploys locally).

I'm really looking forward to the point where I can just type in 1 command and spin up a load balancer, 3 app servers, a cache server and a db server in as long as it takes for the machines to install whatever it needs.

I'm just setting up for outbound right now. Part of the issue is that I have two sites on one server, and while I can send mail from mail@mySite to my email if I use telnet or sendmail manually, I can't send from a file slurp. This ends up resolving to my server login name and my server name, which of course isn't valid and doesn't send anything. I guess that is a good thing.

I think I made a silly mistake in the cname and mx records.

As for the chef/berkshelf stuff and pretty much anything more complicated than local programs that do nothing, dual booting or buying a second machine is probably the way to go. I would love to try out many things, but having VMs running in VMs doesn't sound appealing to me. VMs just aren't cutting it for me anymore.

I really look forward to forgetting how I did anything at all with a server. Hopefully you are taking tons of notes and creating some bash scripts to make your life easier.

I started with taking notes which turned into a bash script and that transformed into using chef because as soon as you have to provision a server twice with some unorganized crippled notes then it's a huge waste of time.

Chef isn't something tied into using a virtual machine locally, it's a higher level abstraction than bash scripts to deal with provisioning a server for any environment you can think of (development, QA, production, etc.).

It also has other perks too like being able to set your server configuration on a central server and then have every machine in your cluster auto-configure themselves at whatever interval you want.

oh ansible is the other one i see lots of people on HN espousing over chef/puppet. no idea if it actually is in fact better

Does it really matter? The whole process is convoluted, slow, and irritating no matter what you use.

i feel your pain, but you're taking the perspective of a non sysadmin. to someone that knows all this **** cold and provisions hundreds of servers, i'd imagine different tools make a big difference.

Can't be easy even for them. I went to the DO IRC to ask for help. The level of trolling was astonishing. This can't be borne from eating easy cakes.

The flip-side is that there are 3m issues for every problem, and there is no way to pin-point anything. As for all the tools. It is like any tool or framework, you sort of need a bloody nose to really appreciate what it is trying to solve, but my problems are your easy things, so perhaps your optimal tool isn't my optimal tool.

It took some ramp up time to understand the basics but I would say it's been anything but slow now. Taking the time to learn how to properly do this was probably the 2nd best tech decision I've made this year.

The best being...?

I read up on VB+vagrant+chef a bit last year but never used it for anything but toy stuff. Spending a week or so of afteroffice time next year on this is very high on my list (as is general networking, SQL-improving and security)...alas PhD first :P

http://aws.amazon.com/ses/

I get that there are reasons for doing it yourself, but if all you want to do is send email this seems way better.

Or am I missing something?

No, there is no sane reason to do it myself.

I'll try this one and report back. It would be pretty okay to receive emails, but I probably don't want to anyways. Yeah, I know I can forward to my gmail or whatever.

Best thing was rails.

I'm only using vagrant with test-kitchen to quickly run tests on the cookbooks. For deploys I'm just installing everything directly on the box/instance but I've only done 1 real deploy and it's on a small site where all of the services are being ran off 1 machine.

I'm hardly an expert but I'm learning and figuring out the modern/popular chef patterns. There's a really awesome podcast called "the food fight show". http://www.youtube.com/channel/UCjcx...F9KO3jw/videos

I recently wrote a backend for a project that included an email server, and I ended up using SES for outbound emails because it eliminates huge swathes of problems and is super straightforward.

Setting up an inbound email server is a interesting exercise though.

Podcast seems great, thanks for linking. I have set up a machine that can be spun up for Prolog development back when I was working through the stuff.

Lol non-open programming languages...couldn't be bothered to actually clone my current environment because it requires setting up license keys for the Prolog (just used SWI instead for a fully free environment should I ever want to port/work on something else Prolog).

p.s.: Are any of you on irc? I wouldn't mind a 2p2-dev channel on freenode

So we had this discussion a little while ago (back in the spring) about pusher vs your own implementation of a websocket. Playing around with this gem https://github.com/websocket-rails/websocket-rails makes it feel much more complete than it used to be. I did some research on websocket security last night, and it seems like while there are a few things to take into account (mostly just ws vs wss and header stuff), it's not super tough to tighten up. What other considerations am I not thinking about? I.e., why wouldn't I just use this myself now that heroku is supporting websockets (I guess that was part of the reason I never looked into it before)?

It doesn't seem super hard to implement even a moderately complicated chat app.

Also, I really want to find an idea I can implement with my designer friend that could potentially take advantage of my angular knowledge and his design skills for a quick MVP. We have about a month to build out some fun projects, just for fun if nothing else. I want to build a 1 page app because non-JS rich apps bore me.

More market research on the food idea revealed that most of what we were trying to do has been done, though hidden behind layers of UX hell. Probably not worth pursuing. Anyone have the next greatest idea they want to share? :P

I never used websocket-rails but I still see no reason not to use it. Although maybe I would consider using nodejs to run the chat server and then just connect to it in the rails app's JS.

Another option would be using live controllers. It polls but honestly who cares. DHH once said basecamp is doing 100k+ requests/minute with a few web workers and a single instance of redis. Maybe it's not the best fit for a chat app but if you want general "semi-real time" updates it seems viable.

It seems like you're going about this the a bit wrong. If others have done something similar is somewhat irrelevant especially since you mention "UX hell". You should really do some focused research (interviews, cold calls) to figure out if the problem people have is actually fixed by these tools (actualy I suspect you might still be stuck identifying a problem as opposed to a "cool feature idea").
I know you have a dev-mindset (as do I) but you should really feel an urge to find out more if only to confirm that you were indeed right and "this problem is essentially solved" (and to improve your finding this stuff out skills).

Obvious/silly examples: plenty of people did search/social networking before Google/Facebook.

And yes, websockets are cool
---

On a very related note I'm wondering how fast I should be able to spin up a hand coded (as opposed to clicked together) typical landing page (about us, learn more, subscribe to newsletter). Including all the admin type stuff I feel like this should be a pretty fast task?

I'm mostly curious because I feel like bundling the skills it takes for this (some webdev/css/js/sql/whathaveu + server setup via heroku and the likes or DIY + some writing for content + mail configuration) would make for an interesting lecture series...I could even imagine doing it a la
0) Some mini Lean Startup intro, brainstorm, come up with idea
1) Use premade tools to build landing page etc (=takes about 1h max)
2) Be able to DIY
As some sort of quick/nondeep intro to programming course.

Over the holidays I came up with the idea of a website focused on resolving board game disputes. It originally started as a live person you could contact to resolve the dispute (call,live support, whatever) but morphed into a social site where people could vote on the correct resolution. Lots of additional functionality you could add if it was popular.

No charge for the idea, but if you build it and become huge you need to throw me a bone of a million dollars or so.

Hipmunk is a great example of being successful just fixing UX hell.

I recall reading an article that said that if there are many players, the problem isn't solved. Apparently there is only room for one dominant company. I think it was a Paul Graham article. But the gist is no competition is bad, one dominator is terrible, and a lot of competition is great since that indicates an unresolved problem.