A few thousand visits a day across 20 or so sites on a server of that magnitude, so it spends the most of its time idle, so I’ve been on the lookout for something else. Shared hosting is a pain in the rear quarters – I’ve been unable to find something that suited me since TextDrive closed down. The many Linux-based VPS’ didn’t appeal to me either, since that usually involves Xen or some other virtualisation-system – and pain.

So I’ve been reading a lot of good stuff about OpenSolaris in general and it’s Zones/Containers technology in particular, so after playing a bit around with it in a virtual machine on my Mac, I decided to try it out for real with a Joyent Accelerator.

All the marketing hyperbole aside (check out the page, it’s kinda hard to see what they’re actually selling), I’ve now got a small server of my own. Or in fact, I have a slice of a very big server. I think the Solaris Zones thing is somewhere between virtualisation and BSD jails. All the customers on this server share the same Kernel and all the essential stuff, but I have root access to play in my own little world. I can reboot my zone without actually rebooting the server – and best of all, its fast.

I’ve moved this site over on the Accellerator, and after figuring out where everything was on the file system, I had no problems setting it up.

I can’t feel a real difference between my Accelerator and my old server when it comes to serving pages quickly, even though the old server had a lot more resources. So that’s positive. And I’m only going to pay half of what I did before. So yay for Accellerators. I can truly recommend trying it out, even if OpenSolaris is a bit different that what you’re used to. I might even grow to like OpenSolaris better.

And goodbye, EasySpeedy. I’d still recommend these guys if you actually need a full server. But since I don’t, it’s nice to know that I’m going to save a bundle on sharing – and it’s even good for the environment :)

This particular passage made me smile:

For such an important piece of software, this is not an ideal situation.

Besides the obvious understatement, the part about it being an important piece of software makes me realise how little i miss it. Because X is no longer important to me. Since I got my Mac, the only thing I run in X11 is Internet Explorer 6.0 and below, using Darwine. So I get to use two pieces of software that should have died long ago, simultaneously.

Why go hatin’ on X, you might ask? Well, five years of full-time Linux desktop use has given me nothing but hate for X.

Because no matter how nice and subdued it might be under the tight wraps of the newest Ubuntu, it’ll use whatever chance it get to jump out and party like it’s 1984.

If you could enumerate the amount of human time that’s been wasted on developing around the limits of this system, on lost productivity from X crashes, on fixing your config files again and again, on trying to get multiple displays, mice with more than three buttons, modern display adapters or just simple keyboards to work, it would amount to many lifetimes.

It is the great exception to the rule that Linux-developers (the OS in general, not the kernel) are more willing than Microsoft to break backwards compatibility to get rid of awful legacy systems and use better solutions.

It is the bleeding ulcer in the belly of each and every Linux installation. And instead of fixing it, Linux and other Unix-based system just pile layer upon layer of extra stuff on top of it to try and work around its flaws.

The brave souls that work on X deserve the greatest respect, but the free software community needs more dictators able to do a Steve Jobs and say “That sucks, lets’ start over“…

One of the things I have noticed as a result of the change, is how many that do not understand the why’s and the how’s of Drupal’s development.

I’ve had to put up with a lot of grumbling about Drupal’s pace of development, our unstable API, even from my new colleagues (I’ve been working full-time with Drupal since January 7th.), who now work with Drupal on a daily basis – you know who you are, curmudgeons ;)

So, this blog post by Moshe Weitzman from 2006 still seems highly relevant (thanks, chx). Truth is, we still don’t know what we’re doing. We still have no road map or master plan. We will probably still see sweeping API changes every now and then.

My point is that you should be glad that this is so, because all these things are the very reasons for Drupal’s success. Let me explain.

I’ve used Drupal as my main CMS for two years now, although I’ve tried it out earlier – at Drupal 4.5. And my, we’ve come a long way since that. It would be disrespectful to say that Drupal 4.5 sucked, but compared to what we have today, it did.

So why do you clamour for API stability? Do you truly wish that we would have sat down at Drupal 4.5 and said that this API is just fine, thank you very much…?

No you don’t. And the same goes for road maps and master plans. It can be a good thing to have plans, but when it comes to a thing that changes as quickly as the Internet does, it is all but useless to plan more than 12 months ahead, because there’s probably a technology right around the corner that will change things – and make your 5-year plan useless.

At Drupal 4.6 or 4.7, nobody could have predicted our current situation accurately – so why are you asking for us to do the same with Drupal 8 or 9?

Community matters

My final point is the community. I think the main reason people find Drupal hard to grok is the community aspect of it. We often describe ourselves as the colloquial “we”. We the Drupal community. We the people.

The thing is, that there is no coherent “we” when it comes to the Drupal community. “We” is a huge schizophrenic choir. A huge smattering of individual, each with their own issues, their own vision, their own itches. Each has an idea of what would be great for the next version of Drupal. And many of us set out to make that idea become reality by doing what we can. Translate stuff, file bugs, help out with the issue queue or write code. Or maybe just advocate Drupal to others, since the more of us there are, the greater chance there is that someone will have the same itch as you – and go scratch it.

So while Dries is a great leader and has plans and visions for Drupal, his voice is not the only one, just the loudest one. And since he has yet to get his own army and secret police to come wake us up at night and make us code what he wants us to, there is never going to be coherent focus in the development process.

And that’s good too. Because even doctors of Computer Science have been know to make mistakes from time to time. So if he was some day to make a mistake, we would not follow him blindly. And I personally think that that’s the way it’s supposed to be.

So the next time you feel like griping, remember that Drupal is what it is, because of all these things, not in spite of them.

First, our company development server at (http://peytz.dk/) where I spend a lot of time messing around with CVS… sigh

-(mih@dev0)-(84/pts/4)-(~/www/drupal5/sites/asdfasdf-dk.drupal5.dev.peytz.dk/)-
-($ history | awk '{a[$2]++}END{for(i in a){print a[i] " " i}}' | sort -rn | head
296 cvs
210 ls
209 cd
40 rm
29 realDiff
29 mv
25 vim
22 wget
20 tar
15 peytzPublish

realDiff and peytzPublish are some of our home brewed utilities, but the rest you should be familiar with. Funny thing that the number of CVS commands exceeds both cd and ls. (I don’t ignore cd and ls in my history as many do, since I like to be able to repeat them just by pressing up.)

Next, my MacBook Pro. I’ve only had it for two weeks, but there’s already plenty of history:

-(mikl@Barak)-(1/ttys002)-(~)-
-($ history | awk '{a[$2]++}END{for(i in a){print a[i] " " i}}' | sort -rn | head
153 cd
131 ls
103 svn
46 ssh
29 rm
23 sudo
22 mv
22 dig
20 vim
20 cp

Not many surprises there. I use SVN for revision control for Drupal Danmark, our Danish Drupal community. Most of my hobby project use git or bzr, but I have not worked with any of them since I got my Mac.

Finally, my web server:

231 ls
210 cd
72 git
64 sudo
55 wget
51 rm
47 bzr
37 vim
32 tar
29 mv

Not many surprises here. Apparently, I have been using git more than bzr lately, probably due to my work on a git mirror for Drupal CVS. This is from a rather long period, since I rarely do much other than log in, do a sudo aptitude and maybe update a few Drupal modules with wget and tar.

For more of these shell history lists, see http://diveintomark.org/archives/2008/04/15/history-meme

In other news, I have started using Markdown for this blog. I have wavered a bit between writing my own HTML, WYSIWYG and Textile for a while, but I decided to give Markdown a try, and I find the syntax a bit more natural to me than Textile. But giving my tendancy to change my mind often, this might not the the final word in the matter…