Jin's Blog

Trying to make the computer do what I want and failing miserably

[jinwoopark]

Web Audio API and Chrome Extensions (with some hacky js hackery on the side)

I was talking to another developer asking me what she needed to do to do some cool audio visualizations with soundcloud.

The first thing I thought of was use the soundcloud API, download the file, run it through an audio analyizer and then do some visuzliations with d3 or some other visualization framework.

She told me that she wanted to do it in real time and as a chrome extension. Hmmmm. I have done some chrome extension stuff before but nothing with audio analysis and visualization.

I looked up web audio visualizations on google and found this from mdn (https://developer.mozilla.org/en-US/docs/Web/API/Web_Audio_API/Visualizations_with_Web_Audio_API) and this http://webaudioapi.com/samples/visualizer/ . The basic idea behind those were to create an audio element, feed it into the audio context from the web audio api and use the web audio analyzer to get frequency data out of it to visualize.

I spent a few hours splunking through the soundcloud js code to figure out where and when they were initializing the audio object and try to hook into it. Using the chrome debugger, I was able to figure it out, but due to (valid) limitations on the chrome extension api, I was not able to connect the js variable of the audio element to the js in the chrome extension.

I looekd online again and found a hacky hack where you monkey patch the createElement method and keep track of all the audio elements created by createElement and then you can access the audio element through the dom. This method required another hack, since chrome extensions don't allow you to run js in the page scope. I created a script element, made a string of js of the createElement hack and then appended it to the script tag. I appended the script tag to the top of the page and then I was on my way.

I cleaned up the code and had a working example, using content scripts to handle the audio and web audio api part and then passing the formatted frequency data to the popup js code where the visualization happens.

This worked for soundcloud, but how about other sites?

Since I had a better idea of what I was looking for at this time, I searched for audio eq chrome extensions (since an EQ would require the same access to the audio as a visualizer) and to my surprise, I found a few. I found one that was on github https://github.com/ejci/Chrome-Audio-EQ and took a look at the code. While it says it only works for only audio and video elements, there was some in progress code https://github.com/ejci/Chrome-Audio-EQ/blob/master/extension/background/background_v2.js which uses something called tabCapture from the chrome api.

From what I found, this is commonly used for webRTC to share your screen for screensharing, but this can also be used to capture all audio coming out of your current tab, which is exactly what I needed.

Since using this api required my code to not be in the content scope but in the background scope. I made modifications to my messaging code but kept most of the business logic the same. I needed to manually close the audio context and the tabCapture stream to be able to the switch between tabs and allow the audio stream to be passed back to original tab.

The code is here https://github.com/jinpark/audio-visualizer and the original super hacky code is in this commit https://github.com/jinpark/audio-visualizer/tree/3401a5ffae729c59b99600f5b9288980f9e3c7fe

Thanks to Soeun for the idea! http://mojosoeun.github.io/

[jinwoopark]

Writing an Apple TVOS app with (TV)JS and TVML

tl;dr version: I made an Apple TV app with atvjs and it was fun and easy.

I listen to a few podcasts and one of my favorites is Accidental Tech Podcast, a podcast run my Marco Arment, Casey Liss and, my favorite host, John Siracusa. They talk about tech and complain about stuff (mostly Apple) but one of the things they love to hate is the new Apple TV, which is sadly is not available in Korea.

I asked my friend from the US who is visiting to bring one over for me and within a day, I had a working TVOS app running on the emulator. Apple took the javascript on the server idea and allowed developers to create very simple apps using prebuilt templates and js for simple event handling, routing and data manipulation.

Looking at TVJS as a modern javascript developer, it is pretty lacking when compared to current tools and frameworks but someone noticed that and created a wrapper framework called atvjs https://github.com/emadalam/atvjs which simplifies a lot of the repetiveness of the vanilla TVML framework.

I already had a rails server that served videos so it was a breeze connecting an API with atvjs. Creating an app that shows seasons and episodes of a show and plays them back. I used handlebars.js to do some small templating.

You can also inspect your app using safari, similar to a phonegap app, which was a nice surprise. Using that with the log helpers in handlebars helped a lot in figuring out templating problems.

A few small complaints that I had was that it was hard to transfer state when navigating and had to create custom handlers to read from the TVML dom to get state. I'm sure there is a better way to do this but I couldn't find a way. Also I got stuck when I saw that JSCore in the Apple TV does not log arrays correctly. If you log a standard array with an object in it such as [{cat: 1}] it shows it as {}. It does not affect the logic at all, but it was surprising.

I had a super fun time developing for the Apple TV and I'm already mostly finished with a second app!

[jinwoopark]

Moving my IRC stack to Docker

Recently, I got a good deal from a cloud provider company called cloudatcost (not linking on purpose here since they seem pretty sketchy) and wanted to move some of my personal services on there.

My current IRC "stack" (inspircd for the irc server, znc as an irc bouncer and common, an irc bot) has been working more or less pretty great with no problems on Digital Ocean for five years or so. But since I'm cheap, wanted a project to do, and I like making my life more difficult, I decided to move it to a cheaper, less reliable and slower host.

The first service I moved over was znc, an IRC bouncer. Since I wanted to do this the easiest way possible, I looked online and found https://github.com/jimeh/docker-znc , a premade dockerfile to run znc. I took my older znc config, made a few slight changes and it was ready to go.

The cool thing about this dockerfile was that it had a way to build znc packages on startup. It just ran znc-buildmod on any *.cpp file it found in a specific folder. Normally, that would have been fine but the module I wanted to use (a module that gave push notifications on palaver) did not work that way, and this specific module required you to run a makefile to get it built. I just ended up exec-ing into the container and running make, and since the module folder is mounted, it would persist the module through restarts.

The second service I moved over was my irc bot, named common, (blessed be based bob). It first ran as a phenny (v2) bot. Then the bot was upgraded to jenni by the developers and then to willie and now sopel http://sopel.chat/.

This was fairly simple to port to docker as well. Made a dockerfile where it installed sopel through pip, created a new sopel user and ran sopel through that. Like znc, it required you to run as a non-root user. I also did a volume mount to share the .cfg files and the modules. The only problem was porting my old, poorly written and documented custom modules to the new version of the bot. Dockerfile at https://github.com/jinpark/sopel-docker

Last but not least was inspircd, the irc server. I saved this one for last because I thought this would be the most complicated. Turned out it was one of the most simple. The only difference here was that I had to pre-compile the inspircd binary beforehand within the docker container. I just did that by docker runing into a bash shell and running configure and make on there while outputting the built output into an attached volume. After that I could just move my config files, run ./inspircd start and we would be good to go.

Overall, moving all my irc services to docker gave me a chance to update my old crufty services while also providing a portable way to change vps providers when the time comes. It was also a lot easier than I expected, most of the troubles were based on the specific software I was using more than the docker stuff. The small docker troubles I ran into (cached images caused apt-get to fail) were things I ran into before when fooling around with docker.

Goooo docker!

UPDATE: The next day after I wrote this post, my vps got completely hosed. Unresponsive customer support and the vps is completely disconnected. Good thing I made my services into docker containers so its super easy to move.

[jinwoopark]

Writing Desktop Apps with Electron

One of my long term 'projects' was to figure out a way for me to look at my todo list more frequently. I spend most of my time on the computer at a browser window, at the terminal or at the IDE, so I needed something that would stay on top of all of those and force me to look at it. The hack I originally had was to make a todo app (Any.do) window transparent and force it above all the other windows. Afloat was what I used, and it required a decent amount of weird hackery and didn't even behave consistently across restarts. While this kinda worked, all the apps enforced minimum window sizes and generally did not fit my needs. Time to keep looking...

Continuing from my last post about ionic, Electron, can be seen as a phonegap for desktop apps. It embeds a nodejs runtime which has connections to native APIs into a desktop app so you can use web technologies to build up your desktop app. Unlike ionic, it does not force you to use a specific framework or give you any native looking styling. It runs a renderer process, which runs the web frontend chromium and a main process, which is the node backend that has access to the native APIs. They talk to each other using IPC and/or RPC. I found menubar on github which allowed you to run mini apps on the menubar and I had an idea. I really like google keep on mobile, I wonder what I can do with it as a native app.

I tried the menubar example app and then quickly modified it to point to keep.google.com and it just worked. I was more or less pretty happy with how it looked and its performance! I created a new icon, made it preload the window so it works a bit faster, added a clean way to quit the app (right click on the icon) and then made it transparent by using some css hackery and stay on top so I can be reminded all the time of what I need to do for the day.

The next test app I made was a pairing timer. I'm starting to teach programming more and I want to bring the joys of pairing to a new group of people. What I wanted here was a pairing app that has a set time (15 minutes), shows a notification and makes a sound when the times up. I ended up using the html5 notifications for mac and tray bubbles for windows and sounds using html5 audio. Most of these I was very familiar with due to it being exactly the same as front end development. Overall, it took a little bit over an hour to code up and have it up and running locally. You can download the mac version here. The source code is here. Feel free to make issues for bugs or feature requests.

The problems I had were mostly at the end. Packaging the app became a pain in the butt, partly because I misunderstood what the packager was doing and the fact that building for windows required wine. Another problem was that for windows, you either need to run the exe file in the same folder as the node runtime or package it into an installer. The problem with the installer is that it requires it to be codesigned, which costs at least 200 usd for a certificate. For now, I'll just create a shortcut and run it that way but we'll see how that goes.

Overall, I had a great experience with electron so far and a great first foray into desktop app development!

[jinwoopark]

Making mobile apps with Ionic

As a developer that mostly works on the web, mobile apps always seemed like it was a big pain to learn. I don't have java or obj-c/swift experience and while I did take an android MOOC course sponsored by google, I came out with more of a basic understanding rather than a practical understanding.

Before that, I fooled around with phonegap, which is a wrapper around webviews of different mobile OS-es. It provides a few plugins to allow interaction with the native APIs, such as the camera, but at the time, I did not feel that it was easy to use.

After that I found ionic. This was pre 1.0 when I did not have much if any angular experience. I explored through the documentation and thought it was great if I could ever figure out how angular works :D

A few years after that, a friend and I decided to start a business focusing on a mobile app. I chose Ionic as the platform I wanted to use and started hacking on it. There were a lot of small inconsistencies with it but it was not too difficult to go around it. I'm still working on this app.

Then I built an app for my friends podcast, Something Sunshine. It was a very simple app, a list view leading to a detail card view. The only problem was getting streaming to work. None of the standard solutions for playing file media worked out of the box and the rest of the streaming solutions were hacky and inconsistent. I ended up using straight html5 audio with an angular wrapper but it is still an incomplete product since it has problems streaming when the screen is off. You can still find the app on ios and android

I then took what I learned from that app and created another app for my friend's wedding. This one is more fully featured, using a side menu with many different views, using the google maps api with directions and accessing the native camera API for taking selfies. I also ended up using the instagram api and used parse as a backend and a poor but manageable admin page. It is also out on ios and android

Overall, I'm pretty happy with Ionic as a mobile development platform. I've hit its limitations a few times but the development team is super quick at answering questions and they have a great development platform with a nice CLI and development app viewers. The only problem I saw was a lack of fully open source large scale apps from which to view. There are many small snippets but nothing that resembles something along the lines of facebook in scope.

[jinwoopark]

"Crunch time" - Writing code for a deadline that has already past!

So continuing from my previous post where I wrote a podcast site using jekyll and markdown and then rewrote the whole site in django...

When creating a podcast site, one of the most important parts is to actually play audio. I looked at a bunch of different radio and podcast sites (http://monocle.com/radio/ and http://serialpodcast.org/ jumped out at me for being better than most) to get some inspiration and found that the majority of them are either using their own custom html5 audio player or using soundcloud. As I've had "problems"* using soundcloud before, I decided to go the html5 audio way.

I went on a google binge trying to find a library that would let me control the styling and experience of html5 audio without me having to do too much work and I came upon amplitude.js. It had great documentation, was fairly easy to use and the authors were very responsive. I asked about a problem on their gitter chat and got a response and fix within minutes! The only custom work I did was to create a range element that was clickable and linked with seeking audio. I have it here if you want to take a look. It was mostly based on a css tricks link page with some css hacks to get the colors to work.

All was well except that once in a while I was unable to play audio until I refreshed and it kept getting worse and worse until nothing played. I notified the library authors about this but they couldn't find a reason why it was failing.

A few days ago, Bobby, Ahnmin and Jackie finished editing the new episode of Something Sunshine and wanted to post it up. They were able to upload and publish an eisode by themselves and everything (almost) worked! The only problem was that none of the episodes played......

Crunch time...

Just to get something working, I hacked it so each button linked to a html5 audio element and just played/paused, without allowing for seeking. This was fine for a temp job but I knew we wanted full functionality before we started promoting.

I spent a day writing a quick jquery library that would give me the same functionality as amplitude.js but actually work. I came up with jquery.audio.js, which is also a quick hack of a jquery method. I'm kinda proud of it (even though the code is horrible) because of how fast I was able to get it out.

Check it out and please send pull requests to clean up the code!

Look at my previous post to figure out what those "problems" were.

[jinwoopark]

Docker and Tutum

Actually Docker, Docker Compose, Circle CI, Docker Hub, Digital Ocean, HAProxy and Tutum

To continue my education of best backend and ops practices, I decided to get better at understanding docker and use it in my own projects.

I've used docker and boot2docker at work but I never fully understood it completely and after one of my coworkers gave a presentation on docker compose, I felt like it was high time that I started using docker for real and reduce my dependance on PAAS providers like heroku.

I followed the instructions on Real Python. One note is when you take the pre compose files, make sure that you delete or edit the .gitignore file since it prevents your changes to circle.yml and docker-compose files from being committed. I was fairly happy to see that the workflow, of using a CI service to test and then, on success, to "deploy" to another service (docker hub in this case), was the same as in my imageresizer service that I've written about previously.

The next step after that tutorial was deploying it onto a VPS. I took a look at the next step that they provided but I did not like the way they handled deployment. It used an api hook to call a shell script to update the container, which I felt was a bit weak. I took a look around and found Tutum. They are slightly abstracting the deployment steps of docker containers. It's not as "fancy" as dokku but I felt it fit my use case. As time goes on I am pretty sure that I'll start handing deployment myself but I want to make sure that I can get it working first with some help.

I changed my imageresizer code to make it run as a docker container here and after going through circle ci, it gets saved as a docker imge here.

After deploying it on dokku and fooling around with exposing the right ports, it was working great. The next step was using some type of reverse proxy to host and expose multiple docker containers on one VPS. Tutum provides a HAProxy container that does this for you "automagically". Except that the magic part needs to be worked on a bit.

I spent some time figuring out how to get this working after following the instructions. What I had to do to get this working was to add a VIRTUAL_HOST env variable within each of my services with the url of what I wanted to route, and also change the BACKEND_PORTS env on the haproxy service to what the linked ports are for your services. I think Tutum is supposed to handle this automatically but I don't see this happening.

You can see this working by going to http://images1.jinpark.net/http://i.imgur.com/6GSvBbn.jpg/?rwidth=200 which is the image resizer service and at http://hello.jinpark.net which is the standard hello world tutum container. Both these containers are hosted on the same Digital Ocean VPS and are routed through HAProxy which was handled by Tutum.

[jinwoopark]

Creating an image microservice with flask, imagemagick and cloudflare

At work, we have a microservice that allows for cropping, resizing and caching of images. This allows us to use the same assets that a designer would create and resize it as needed. It also allows for edge caching through a CDN for users to get the images faster. A very useful service.

I saw firesize on product hunt recently and I remembered that I wanted to remake our image service on my own, for fun and non-profit. The last project I made was with ruby and sinatra, so I decided to (very slightly) switch it up and use python and flask.

I decided on using imagemagick for the imaging library because it was easily installed (or preinstalled) on heroku and had a great, simple python api (wand).

Currently, I just have it resizing based on width and/or height and converting based on filetype. It uses a url scheme of someting like

images.jinpark.net/image_url?rwidth=optional_width_in_px&rheight=option_height_in_px&type=optional_file_format_extension*

To include caching, I used cloudflare's free tier to allow for edge caching.

Theres some more work to be done, like adding cropping and maybe even seam carving. The biggest change would be to add async processes so requests wont be blocked.

The github repo is here and in the repo, you can deploy to heroku with a button!

[jinwoopark]

Using jekyll/github-pages to host a podcast site

Two of my friends (Ahnmin and Bobby) at work started making a podcast together called Something Sunshine and like any podcast, they needed a website. Since this was just a fun project for them, I wanted to make sure that they did not have to pay for anything.

I was thinking of what would be easy and free and my first thought went to meteor, a new js framework that I was learning. It reminded me of rails a lot due to its convention over configuration mindset and its ease of use. I started setting that up but realized that I might be over engineering this. What they needed was a simple cms where they can upload a "post" or a podcast episode and some metadata. Meteor was for a full application. Definitely not the right usecase. Time to think of a simpler alternative.

At work (DramaFever), we are trying to move our flatpages (embedable html pages usually used for marketing), into github so there could be at least some semblance of a code review when our designers upload their new designs.

I wanted to take that same idea and use it for the podcast site.

Jekyll is a very simple blog platform written in ruby that is hostable (for free) on github pages that uses simple markdown files as blog posts. It also allows simple metadata using YAML's front matter. It's not as nice as a full CMS like wordpress, but its very lightweight and easy to learn.

I set up a simple site using fullpage.js and, since the podcast authors wanted to use soundcloud, a very simple soundcloud embed system with a few event handlers so we can keep track of some basic stats through google analytics. I also saw that others are using jekyll for podcasting and have even made itunes compatible rss feeds through jekylls templating service.

I haven't opensourced it yet due to it being in progress and theres a lot of application specific things that are in the code. After I pull that out into its own settings file, it'll be out in the wild for anyone to use :)

Seriously, check out Something Sunshine. It's a podcast about movies, video games, music and anything else that comes to mind. They are great hosts and extremely fun to listen to!

As an update (May 23 2015), We ended up moving to django and using their admin to handle everything. This was more of a monetary thing than a tech thing. We were using soundcloud to host our files but were limited to a 3 hour max limit for the free plan. We could have moved it to the repo itself but that would have been fairly complicated for our non technical podcast makers. I've open sourced the repo but its pretty dirty and I haven't gotten around to cleaning it up

[jinwoopark]

Setting up a dev environment for wordpress

This might be extremely boring for anyone who maintains servers as it is the most basic level stuff that you guys probably know :P

I've been helping a friend monitor and update her wordpress blog that gets a decent amount of traffic.

She wanted to do some large changes to her blog but didn't want any downtime. So, the next step was to get a quick dev environment up so that she could test her changes and get a feel of how things changed.

Cloudatcost is a Canada based provider of cheap VPS services. They have a $35/lifetime plan for a low end VPS that I'm using. Reviews say they suck, especially regarding uptime, but for something like this where uptime is not important. I feel like this was the right choice.

I installed ubuntu 14.04 x64 and started using this digital ocean tutorial to get started. I was using this plugin. This plugin needed apache/php/mysql installed but not wordpress.

So after everything was installed on the dev server and the backup/installation files were copied over to the html folder on the machine. Before you install, you should modify the permissions of the html folder so php could write to it. After that's done, you also need to enable mod_rewrite with this command

sudo a2enmod rewrite

add

<Directory /var/www/html> Options Indexes FollowSymLinks MultiViews AllowOverride All Order allow,deny allow from all </Directory>

to your /etc/apache2/sites-available/000-default.conf file and then restart apache with

sudo service apache2 restart

This allows the .htaccess file to be created by wordpress so you can modify your permalinks. Most of this is written here

Run through the installation and check your wp-config.php file to check for references to your old domain. I had to remove the caching plugin and reenable to get it to work.

Now everything should work!

[jinwoopark]

Wechat Sticker Maker

One of my friends visited me in Korea and she got me hooked into Wechat, another messaging platform. The cool thing is that it allows you to use custom gifs as stickers! (Yes, I know there are other messaging platforms that do this)

I found this tutorial online (https://www.techinasia.com/favorite-animated-gifs-custom-stickers-wechat/) which pretty much comes down to two steps.

Resize the image to between 150 - 200 px

Optimize the image so it stays under 250 kb

After that, you can save it to wechat and send it to all your friends.

At first I was doing this "manually" using some gif optimizing tools I found online but then I realized that I could do this myself and save myself the hassle of choosing how I want it to be optimized.

I saw that the gif optimizing tool I was using was using gifsicle (http://www.lcdf.org/gifsicle/) in the backend to do the resizing and optimizing so I used a node port of gifsicle (https://github.com/imagemin/gifsicle-bin) and set it to resize and optimize the image until it comes under 200kb. Sometimes this is not possible with huge or long gifs so it returns an error when that happens.

This is hosted on heroku and has a very simple (read ugly) web interface that lets you upload a gif or send in a url of a gif and it will resize and optimize it and return it to you so you can save it.

It was a super quick weekend hack that I'm pretty happy with. It does what I want in a quick way. Lots to do if I want to optimize it but I'm ok with it as is for what it does.

http://wechat.jinpark.net

[jinwoopark]

Analysis of irc logs

I've been on IRC for a while (5 years or so).

I made a lot of friends and talked to a lot of different people and watched myself and others change over time.

Recently, vertigo (one of the early members of the IRC chan I was/am in) got me ~7 years of logs, which came out to be ~2 million lines.

I loaded it into mongodb (so I can get more used to nosql), installed pymongo and decided to do some sentiment analysis on it. I used the great TextBlob module to quickly do the sentiment analysis.

There were a few things that vertigo and I wanted to find out.

One was who is the most positive and negative in the channel. I did some VERY basic analysis and looked at the average sentiment polarity and got some interesting and some obvious results.

complex (used to be) our resident troll-ish person. Worked perfectly that he was the most negative. Another user PF used both lowercase and uppercase versions of his nick and we can see that both of them are almost the same.

Another question was since vertigo gets some "help" from certain substances during the weekend, is he more positive during the weekend compared the weekdays.

As we can see, there is a slight trend but it does not seem to be significant. Also vertigo likes tuesdays for some reason.

Next step is to look at some time series and see how people changed over time!

[jinwoopark]

2013-01-30

2013-01-30

Yesterday I went to a meetup called Obtaining, Scrubbing, and Exploring Data at the Command Line

It was just a mix of old and new things. The standard unix command line tools (cat, awk, grep, sed, less, head, tail, etc) and a mix of new tools (csvkit - a bunch of small command line tools for csv, jq - tools for handling json on the command line, xml2json and json2csv and others)

The speaker also got into a bit of his own tools. From something big like being able to call R from the command line to small bash functions that uses http://explainshell.com to print out what a shell function does on the command line.

He talked about creating "data science toolkits" and gave some examples of packaging and creating enviroments that can be easily moved around. He has his up on http://datasciencetoolbox.com

I took some notes but someone else seems to have done a better job then me. He has posted his notes here.

[jinwoopark]

Sending text message on a schedule with twilio and at

My friend Anna is in Canada right now. Canada has delicious things that America does not have and I would like Anna to bring me some.

I gave her a small list of things but she said she would forget and would prefer for me to text her what I wanted Thursday morning (which is when she would have time to buy stuff). That is a few days away and I was sure I would not remember when Thursday rolled along. So I decided to set up a script somehow to send a text message to Anna on Thursday morning. I had a server running, a twilio account and I knew linux would have some sort of command to set up something time based (most likely cron).

I had a phone number on Twilio before when I used to host a text message service for App Academy friends on who they would be paired with the next day. (I will probably write a small blog post on that later)

Setting up the ruby script was very simple. I just used the ruby example on Twilio docs and changed some of the config values. I tested it by sending a message to myself and it worked perfectly.

I knew I wanted to set something up to shoot off this script at a specific time. I knew that cron would work but when I googled, it seemed that the "at" command was a much better choice. Cron was made for repeated schedules while "at" was for one-offs. Running it was extremely simple. To send something off at a specific time, it was just

at 7:00 am

Or even

at now + 2 min

Awesome simple syntax. There are more examples here. To check out all scheduled "at" jobs. It was just

at -l

So I just set up

ruby <location of file> | at 9:00 am Dec 26 #which is thursday

and checked it with

at -l

and it was done. Woot!

[jinwoopark]

Setting up Primecoin mining on Ubuntu

I had a few credits at a VPS that I wasn't using so I thought that I could start mining some primecoins.

Setup is pretty easy and I'm using beeeer as a mining pool because it seems like solo mining is not worth it anymore. I mostly followed the instructions on here but I also added

apt-get install -y libboost1.48-all libboost-chrono1.48-dev

before running the makefile. Also, if your build is failing, it might be due to the lack of ram. To add a swapfile,

sudo dd if=/dev/zero of=/swapfile bs=64M count=16 sudo mkswap /swapfile sudo swapon /swapfile

Then

cd ~/primecoin/src make -f makefile.unix

and now you should have primeminer as an executable.

usage: ./primeminer -poolfee=<percentage> -poolip=54.200.248.75 -poolport=1337 -pooluser=<address> -poolpassword=PASSWORD -genproclimit=<number of threads>

[jinwoopark]

Setting up shairport on raspberry pi

Overall, it went fairly well besides for the fact that the SD card I was writing to seems to not want to work with the rpi and the wireless usb dongles I'm using seem to be too weak to work well.

Well, whatever.

I mostly followed instructions on http://jordanburgess.com/post/38986434391/raspberry-pi-airplay

and complemented by the wifi help on http://jai-tech.blogspot.com/2013/03/setting-up-wifi-on-raspberry-pi.html

Theres a new version of shairport that I should try but I'm pretty sure the main problem is the weak signal from the router to the rpi. Not sure how to fix that. I should probably get a new router.

[jinwoopark]

Setting up inspircd

I'm going to try to write more and more on this blog because, why not?

Today and yesterday, a bunch of us on our irc channel decided to move our channel to our own network because we were scared of the stability and of the alertness of the opers.

I first installed it on my mini server to figure out what to do and then installed it on my digital ocean vps.

I already had a hybrid-ircd on the server for my app academy friends and I didn't want to kick them out and move them before I had a working server so I just built it to work side by side. (Note: standard apt-get install for inspircd killed hybrid-ircd so make sure you build from source) I mostly followed the sparse instructions from http://www.inspircd.org/wiki/Installation/Source.html

If you want to use ssl, install openssl and libssl-devel beforehand and when you do ./config it will pick them up. Also run

openssl req -x509 -nodes -newkey rsa:1024 -keyout key.pem -out cert.pem

in your config folder to create new certs.

The only part you have to pay attention to is the common name. Make sure you put that as your url that you are using.

After ./config and make and make install. Go to your inspircd.conf file and make your necessary modifications. Then change your links.conf to be able to connect to other servers and oper.conf to allow yourself to be an operator.

I will add more things as time goes on.