Lol I'll most certainly use a default theme. Looked through them and like Kubrick best, maybe if I'd find one that is green where that is blue I'd use that.

If I ever get to constant posting I'll worry about themes

I use a default theme on my Drupal blog. I did, however, change the header color, so that's cool. I get like 30 unique visitors a month.

If I was seriously adventurous, I'd use this Clojure Blog:

https://github.com/ibdknox/Noir-blog

But if feels like a waste of time to commit all that energy to something I would hardly use, people would hardly see, and I'd never be able to sell to someone looking to build a blog. Going nerd for the sake of nerd is cool and all, but in this case, it's impractical. Maybe if I get 10,000 unique visitors a month, I'll reconsider.

I wrote a Git backed NodeJS blog a few months back. I dont blog so actively any more, but there are a few people out there using it

https://github.com/guyht/Glog

there are plenty of free themes that work fine, no need to get too crazy. as everyone knows, content matters most

I saw that you already went another route, but have you used AWS before? Everyone gets a free year of a micro EC2 instance. If you've got APC installed and use InnoDB with innodb_flush_log_at_trx_commit = 2, it should probably handle a decent amount of traffic fairly easily.

Speaking of which: I am on the home stretch of this website. Getting the program to talk database probably shouldn't feel this good, but now that I have it, I'm like:



Actually, this is sort of the beginning: the entire (planned) functionality is pretty database-intensive.

Sadly, I'm already thinking about how much I would like to rewrite the entire thing. Is this a normal feeling as a function that, the more you learn, the more you wish it was done some other way?

Standard.

Fight the urge! That way lies second system syndrome.

Yeah...you just look back, shake your head at yourself and then do the next one better (lather, rinse, repeat ad infinitum)

http://c2.com/cgi/wiki?SecondSystemEffect

One comment has the perfect solution:

I have a better solution: Stop learning, thus every time you approach a problem, you will be very careful and treat it as your first system. If that's too hard, just use an entirely new programming language for every project you build.

Wow, worst advice ever maybe? Here's a tip, you learn by iterating. You don't learn by doing the same garbage code over and over.

Btw _dave_, can you perma. ban me from this sub-forum? I came to the conclusion that nothing good comes from viewing and posting here.

Didn't Brooks also say you should plan on building two and throwing the first away? Now he doesn't want you to build the second one and just move on to the third? Maybe that's why IBM is such a powerhouse today...

I think his point was that you don't totally rebuild something that works just for the sake of making it better. Obviously the next time you do something you do it better.

Edit: http://www.joelonsoftware.com/articl...000000069.html I don't totally agree with this blog post - but I think there's some valid points in there.

Seconded.

I considered linking that but I didn't feel like adding disclaimers. Joel, especially in the old days, wrote from a very particular perspective of somebody writing software as a product in the Microsoft model. So you had to translate.

Lol, definitely.

Basically 90% of blog posts that Joel wrote I was like:

And 10% were just pure trash.

Any recommendations on learning Ruby/RWR besides searching stackoverflow

I don't know what RWR is (Ruby's so faddish, unfortunately), but for learning Ruby... it really depends on what language you're coming from.

Coming mostly from C++, some python and no Perl.

Meant Ruby / Ruby on Rails

Sent you a PM.

Ah. Well learn Ruby first before any particular library. And the adjustment from C++ OO to Ruby OO is going to be a thing.

This chart is biased but it highlights differences

I know books are generally knocked here for good reason, but the Ruby book by the Pragmatic Programmers may have value for you, introducing you to a Ruby way of thinking. I'd point you to the why thing if he hadn't gone insane before finishing and polishing it.

I got a question regarding monetizing unique visitors.

Let’s assume I have a website that gets unique visitors each day.
Those visitors are unlikely to click on the ads a lot and the chance of them clicking through and buying something are very, very low (let’s assume it’s 0%).
The visitors will be repeating and the content of the website is of some use to them. The content will change daily, so they got a reason to return every day.
Given there is a search on the website, they will click on more than 1 page on each visit.

The problem with the site is, the marginal appeal of it is negative and the most traffic will be at the beginning and then be more sporadic on an ongoing basis.

In addition, the content is easily replicable, so I assume someone to replicate it rather sooner than later. I am not restricted on how fast I have to execute. Although, it should be rather easy after I have found the right ad provider. The website is easily done within a few days.


I want to maximize the $/unique visit given the aforementioned “restraints”.

Is there an optimal ad network to use? I want to have as minimal interaction and hassle with them as possible.

I don’t know how well Google ads would work, given my user base.


Anyone with suggestions?


Also, in addition to this, I got a few questions regarding the website.
I want to do it in wordpress, it will literally be two ad banners, a content output and maybe a menu.
The content would be filled with datasets from SQL queries, how easy is this doable with Wordpress? I got the SQL queries (at least the MS Access versions of it, which I assume to be almost identical).

My database only exists in Access at the moment. I got over a million datasets in there already and I expect it to grow into smallish 8 figures. Which database format (or w/e it's called) would you recommend?

Following up on this, I want to fill the online database with the datasets I have at the moment. I can get them in CSV format easily if required. I basically want to mass copy those 1.x million datasets all at once. What's the best way of doing it?

And a somewhat related question: I want to datamine using an online server and online script. Is this difficult and which programming language would be best to use? I would exclusively parse through sourcecode.


Thanks in advance, might be a bit long. I can create a seperate thread if necessary.

The way I interpreted is: "Don't rewrite something that already functions just so you can feel better. Sure refactor, create extra namespaces (don't ask), and clean up what you already have, but don't rewrite the same friggin' program for ego purposes."

Considering the generally up/down learning curves we all experience, I can understand the argument. There is something to be said about doing things by the book, but humans tend to think that once the book stuff is understood, then it is time to break some rules, and the danger is that a little experience tends to create over-confidence, thus you end up breaking all of the rules, and that is not good. I guess the poker correlation is you start up playing by the book (TAG), then you think you know all the rules, so you loosen up (LAG), then realize you don't know all the rules to break, so you go back to the book (TAG+ Bruised), then reiterate.

I was looking at my code-base and it is so obvious what parts were written first and what parts were written second. I already did a ton of refactoring. The example I posted earlier where I used string-args was just awful, but at least that and similar stuff is gone. I just see a ton of little things that irritate me, that's all, and it's not hard to convince yourself that rebuilding from scratch would be easier than maintaining or upgrading.

With that said, I think it would be a good idea to add new namespaces to clarify the organization of the code more. I get very nervous when I see files that exceed 100 LOC (that roughly translates to 3,000 LOC of Java, LOL). It also doesn't help that I am using the knowledge base gleaned from about 350 pages + using a programming language that I'm not familiar with at all. The good news is that I bought Programming Clojure the other day and that helped me to clarify some meta-design issues I was undecided about.

I enjoy reading your posts: you encapsulate the balls-to-the-wall hacker spirit, and there is always lively debate. I give you tons of credit for turning jQuery into a meme .

I suppose that, if you were to refactor, add new features, and fix code as you go along, the base you have in 5 years wouldn't look completely like what you have today, right?

The framework I am using has a built-in SQL security function. It's interesting because it hashes and then salts the input to to max varchar limit. One side of me thinks this is cool, so I don't have to write my own salt and string-escape functions, but the other side was irritated because I had to alter my tables and there's nothing I hate worse than altering my tables.

Security Question: Is a variable salt maxed to 255 varchars a little bit overkill? What is the minimum, and how far over minimum is this?

Then I head off into other wild territory: now that I don't trust the user to create a good randomized password, I also don't trust them to use unique passwords and usernames for each site they are on, so now I have to encrypt the usernames now, and of course, if their email was compromised and they use the same pw on the emial, I have to now encrypt the email, and oh wait! I can't stop there, I have to encrypt the phone numbers, and the addresses, and the names, and the....

So, how much encryption is overkill, and where do you stop without turning the entire database into an maxed varchar cluster-****?

I don't think length of salt is that important beyond say a handful of chars, but that's a bit of an assumption and I might be wrong.

The length of the salt usually wouldn't have an impact on the length of the hash either, the hash would output at a fixed length (32bit for example)

Also hashing and encryption are different things. Hashing peoples usernames is an interesting idea, but if you salt them there's no way to easily tell if username x is taken yet or not.

I poked around, and apparently I am sort of lucky. The framework uses bcrypt, which appears to be the most respected thing out there. The shocking thing I'm seeing is that many people seem to disavow salting in general, meaning that if your using MD5, et.al. the salt won't help at all with encryption. bcrypt is built on Blowfish, which uses a slowing algorithm to make it too computationally expensive to crack the data. This all sounds nice, but I don't get why anyone can be compromised so often if it is apparently this easy, but screw it, this is easy 10000000X better than rolling my own.

To be certain, I'm not disavowing Java's safe password utilities.

edit to add: I was of course wrong about the salting. I just saw a hash and didn't know what that was.