Brian Bommarito: Programming, Life, Stuff

I mentioned way back in my first post that I host this site with TigerTech, based on Matt Cutt’s recommendation (Or more like a throw away comment on who his site is hosted through). TigerTech is fairly inexpensive, though they use that awful ‘unlimited bandwidth’ moniker on their site (Though after reading through their faq they at least explain it’s not unlimited). Since I started it, hadn’t really noticed any issues until the other day.

I came to the site to check on a few settings, and found access horridly slow. I mean, minute long load times to go past a white screen slow. So slow that I think a snail passed by it on the information superhighway. I don’t know if I can come up with any more analogies to how slow it was, but needless to say it was slow.

I contacted them, secretly hoping it wasn’t my site causing the slow down. Turns out they ad noticed a high load on my server and were investigating it. Pretty standard answer to problems like this from a shared host, but it was what they said after that that surprised me: ‘For the inconvenience, you will not have to pay the July or August fee’. Wait a minute, hold on…2 months free for a slow down on the site? That is pretty well unheard for a site that costs me 7.95 a month. What shocked me further though is they contacted me a couple of times after I contacted them to keep me up on what was going on.

I can now say that TigerTech is a great company, and really worth what you pay. They have done something no other host has ever done: Impress me by being up front, and proactive. Check them out.

I have to ask myself that question rather regularly. Databases can do an awful lot of stuff, including dealing with data integrity, some level of validations (Mostly that a field exists, that two fields are unique, etc, etc) and can rollback if there is a problem with a query. The problem, it seems, is that the people behind Ruby on Rails (And more specifically ActiveRecord) are truly afraid of the database, or barring that, just think of it as a place to stick data; an upgraded file system if you will.

Even with ActiveRecord 3, there have been a ton of changes: A much nicer API, lazy loading of database queries (IOW it only loads from the database when it’s required), and I’m sure a lot of tweaks under the hood, but still no native support for foreign key relationships, still AR needs to scan the database every time to find the columns, and still the need to subclass from it, vs just including it (I hate that; subclassing where it makes no sense is stupid). Oh and if you want a read/write master/slave setup, well forget it, that’s not there either.

This has been the one sticking point for me for the years I have worked with Rails, and ActiveRecord. I come from the world of enterprise development prior to Ruby on Rails. I used to handcode all the SQL queries, I used to take advantage of relationships, and views, and stored procedures, and I still would like to. In fact, stored procedures are a huge step towards data integrity, you only allow certain types of data in, and then the database takes care of the rest (Depending on how you program it). And views/materialized views are a huge help towards dealing with database server load.

So, what can one do if they want these features in AR? Well, there are some gems that sort of help. One to help out with building foreign keys from the migrations, one that helps do multiple databases, but those gems are few and far between.  Even the alternative ORMs out there for Ruby do not really address the more enterprise features, probably because they came out of a set of desires, and enterprise was not one of them. As I see it, there are only two other things that can be done: Bitch about it and hope it appears in another version of AR, or someone takes care of it or do something about it myself. I choose the latter.

This project has no name as of this moment, but I want to build a full ORM system for Ruby that will support what AR supports now (Lazy loading since I love lazy loading, a nice API), and that will offer support for some levels of clustering, as well as the more enterprisey features. Granted, this project is no small task, but I finally decided I want to delve into it just to further my knowledge of Ruby, and to maybe solve my own desires.

I haven’t really talked about RubyNet since I started thinking about it, and planning it about 2 years or so ago. I mainly didn’t speak about it because I hate the “Oh, I’m building X, it’s going to change the world” type posts you see all over the place for various and sundry sites. No, your site will probably not change the world, and not everyone wants to know about it. I have made some vague mentions of it as a project I was putzing with, and that was about it.

I figured I would talk about it today only because the reason that it got stuck in my head still puzzles me, and I have to articulate it just so that maybe it starts making sense to me. The whole idea was sort of started by a friend of mine, who is also a recruiter for RubyFocus. One day we were talking, and he mentioned how hard it is to find Ruby developers, and that struck me as kind of odd.

Ruby is not really an academic unknown language. Yes, Java has more recognition than does Ruby, as does the whole .Net framework, but Ruby is also not Lisp or Smalltalk, which are in production, but most do not know about them, or they know about them, but only in the history of computing. I don’t see Ruby like that, especially with the Ruby on Rails framework being as popular as it is.

It was, however, after that discussion that RubyNet was sort of formed. There are other sites to find Rubyists, including WorkingWithRails, and RubyNow but those are either under-used, or just out of date in terms of technology/userbase. Linked-In isn’t a bad resource, but then you either have to hope the person is in your network, or you can go through the layers of networks until you get in touch with them, and there the accomplishments can’t really be demonstrated (Yes, you can provide links but links don’t tell the whole story). You can also just throw an ad up on Craigslist and hope, but from what I have heard many of those get resumes from people who have never touched Ruby before. Dice.com…forget it, it’s loaded down with recruiters (I get 5-6 contacts a day from Dice.com, all recruiters I had never heard of before).

RubyNet aims to solve those problems. RubyNet is a social network, so picture Facebook for Rubyists, but I think it offers far more functionality than does Facebook. Some of the other features that are currently either planned, or developed as of this moment are:

• Job posting. Yes, I know there are tons of places that this exists, but hey one more doesn’t hurt and if RubyNet can acheive any kind of critical mass, it might be a good place to look.
• Resume hosting. Again, tons of places out there, but I also want to enable searching for skills/years of use of a technology. I think being able to search with a very fine granularity would be a huge benefit.
• Rankings. WorkingWithRails offers this, but since WWR is under-used, the rankings really mean nothing. Besides, the rankings on RubyNet also take into account commits into projects, etc (Integration into Github for now).
• A Mechanical Turk like system. Mechanical Turk is generally used for small projects, but I want to offer a similar service for larger projects as well. You find someone interested in building x or y, all the paperwork is done digitally and through the site, then the money is held in escrow until completion and it’s then paid out. Basically one stop shop to find freelancers.
• Event/Group listings. A given I suppose, but the ability for groups and event planners to list their events or groups, sell tickets (Yes, I plan to offer ticket sales as well for an event) and get the word out about whatever the event or group is about.

Could you do this through Facebook? Well, some of it certainly, some of it not. Though I think somewhere for the Ruby community to congregate, and seek each other out is far more valuable than trying to adhoc something. Besides, unlike Facebook everything in RubyNet is privacy focused, down to encrypting content that is not publicly available so if something does go wrong, that content is never exposed in any meaningful way to a search engine.

RubyNet has no set launch date. It’s just me working on it currently, and I have a day to day job so the work is sporadic at best. I have contemplated seeking some sort of funding, but I haven’t really decided where to start nor who would even be remotely interested in funding such a thing. I’m sure there is someone out there that would be, and if someone reads this and is interested, contact me.