17 days ago
As I mentioned, I had the idea for CFFSW some six months ago. Part of what took me so long to make any progress was a giant amount of faffing about that can only come from knowing enough to be dangerous. In the end, I settled on Wordpress as a platform, knowing it was the one that I would be least tempted to divert my time from building the content to building the platform.
There are a few reasons for that. The first and most obvious being that Wordpress is extremely mature, thus making it far less likely that I should ever need to “go under the hood”. A second one is that it’s PHP, a language I have no desire to develop any competency in.
Something that made me lean away from Wordpress is my experience a couple of years ago, when Wordpress installs were getting exploited left right and centre (especially on Dreamhost). Trying to clean those up was a painful experience, but Dreamhost have made it a lot easier to upgrade Wordpress installs almost immediately (I either get an email with a link to click, or I get an email saying they’ve done it for me) which helps. I tidied up my Dreamhost practices a bit by having only one domain per user (here I mean unix user, I typically only deal with these when installing new things, thereafter most things have a web interface) and also enabling ‘Enhanced Security’ for each of those users (it boggled my mind to learn that this was not the case already). Although Dreamhost claim to enable this by default now, if you make a new user at the time of making a new domain (probably the most common use case) it’s not enabled. So much for that. I digress…
My first iteration of CFFSW actually involved a concerted effort in investigating various Python static-website generator options, and I even chose one (Nikola) and built a first draft. I was quite happy with, it was feature-enough-ful, looked shiny and new enough (thanks Bootstrap), the author/mailing list was responsive, I could write entries in Textile markup (same as Textpattern, which this blog uses) and hey, static sites do not get hacked! I thought I was golden.
But then… I put it down for four months, and when I picked it back up, a huge amount of development had passed. I discovered it is in fact possible for a project to be too active (or maybe, not yet stable enough).
And I remembered how annoying it is to install stuff from source on every computer I use in order to update a blog, which turns out to be at least four. Web interfaces do have their convenience.
However, I have an idea (which I haven’t tried out yet). Github lets you edit/add new files via the web interface, and they do a formatting preview for Textile and other markups. So I can basically use Github as my web interface to updating my blog. I just need to set up some mechanism that rebuilds/reuploads the pages when new commits arrive. (I can accept that requiring a src clone, because it should rarely need updating.) NB: if anyone can point me to some scripts/projects along this line, please do!
Finally, why not more Textpattern? I am quite used to it, but there are several factors against it:
- Security: Requires more work to update, which means I’m more likely to leave it out of date for longer (FTPing files, Wordpress has spoiled me)
- Far harder to change themes (which is why this blog still looks how I felt c2008)
If I use a Python static site generator I can do a little platform-building when I feel the urge. It never goes away completely. :)
21 days ago
So as part of the lightning talks at LCA in January, I decided to launch a little idea of mine: A catalogue (taking form as a blog) called Crowdfunded free software, aiming to record the use of crowdfunding campaigns such as Kickstarter to fund free software and related endeavours. I hope that it will be useful as both a historical record and as a resource for those who want to support free software in this way (ie to find out about new campaigns while they are still open).
I had the idea for it at PyConAU (6 months ago… ahem) when I heard someone casually mention how there had been a Kickstarter campaign to integrate South (library database schema migrations) into Django core. “How come I never heard about this?” It seemed like exactly the kind of thing I would be interested in hearing about. And as we know, when something doesn’t exist in open source, you have to scratch your own itch. But it took me a while to get my act together.
There are still a couple of dozen significant campaigns I would like to write about, and I haven’t completely figured out how to handle new campaigns, small campaigns and very poorly supported campaigns. But in the meantime, I would really like people to let me know when they hear about new free software campaigns! Either via the blog or via Twitter.
(Sidebar: the twitter name cffsw is taken although unused. This, and twitter username character limits, forced me to think of something else, which eventually culminated in crowdfundfloss. I now think this is probably a far better name than the original, which yet again shows that Naming Things Is Hard.)
A brief interview with me conducted by the lovely Onno VK6FLAB :
29 days ago
I joined Newsblur in November 2011, which was just after the sharing functionality in Google Reader was removed. I jumped on board with a paid account and all, even though sharing items was not yet available in Newsblur. At least “social” was in the works, unlike Google Reader, where a slow dismantling was just beginning.
In July 2012 Newblur’s “blurblogs” arrived, and I was content because now I had a page of shared items. It’s not perfect (not a fan of the infinite scroll, for one…), but it’s there, it has RSS, there’s a bookmarklet, it has comments, tick. So if you use Newsblur you can subscribe to that by clicking “view on Newsblur” under my username, or subscribe to the RSS feed in some other reader by following this URL (or probably just point the page at your reader, maybe it has autodetect magic).
here is my one-liner for making your Newsblur shared items page 200% more awesome:
This goes under “Custom CSS for your Blurblog”:
Here Newsblur gives you an idea of the kind of things I like to share:
That screenshot is actually from the intelligence trainer. They have been part of Newsblur from the very start, but I only started using them relatively recently. I was a bit suspicious about how “intelligent” they were going to be. I generally prefer to see everything listed and choose myself what I care about, rather than have software selectively show me things it thinks I want to see. Happily Newsblur takes a simpler approach that works fantastically well, giving you a great deal of control while not overstepping its bounds. Facebook and Twitter could take note.
The training is per feed, and is a simple thumbs-up/neutral/thumbs-down on phrases in titles, tags, and the whole feed. Items then end up being green, neutral (grey/yellow) or red (hidden by default). You can view just green or green + neutral.
Training is useful if you like almost everything in a blog, there’s just one or two topics you don’t care about:
It also works wonderfully well for large sites/busy feeds, where you don’t care about 99% of the content but there is 1% you do care about. Now it would be nice if such sites had custom feeds (eg per tag), but I have almost never seen this on large sites. Even tech news sites. Here is how to fix that:
- Thumbs-down for the whole feed
- Thumbs-up for the topic you care about
(That’s from We Work Remotely… I can’t help noting it’s not as good as Stack Overflow Careers, which has feeds per search)
The one above is from xoJane. They don’t use tags/categories in a granular enough fashion to be useful for this purpose. (One of their categories is “Issues”…uh, ok.) But they do use post titles consistently enough to serve as pseudo-categories. So when I am going through my feed, if I see a type of post (like, open thread) that I never want to see again, I right-click on the title, select all/part of it, click thumbs-down, done.
One thing this means is that you probably don’t ever want to do thumbs-up for a whole site, because if you then thumbs-down a particular tag or title, those items will come out neutral (and depending on your settings, you will still read them) rather than red (never see).
Also, in a somewhat voyeuristic fashion, you can see what kind of training other Newsblur users have done on the same feeds! It’s called “Site statistics” (it’s also available in the API):
A couple of other users agree with me, that open threads are boring. :)
Finally, the Newsblur Android app is excellent. You can do training on it (on tags only, not titles). You can switch from feed view to “text view” to get around annoying partial-content feeds without leaving the app. I almost-maybe-even like using it more than the desktop (browser) app, and there is no other desktop-first app that I feel that way about.
There is not many pieces of software that excite me enough for me to actively evangelise them. For reals, you should try Newsblur.
31 days ago
35 days ago
A month ago I attended my seventh Linux.conf.au, LCA2014 in the lovely Perth. January rushes by so fast I often neglect to write about it at all, but this time I would like to highlight my two favourite talks.
The first is Friday’s keynote, “Deploying software updates to ArduSat in orbit” by Jon Oxer. I would call this talk: “A school class can do a science experiment in space for a week for $300”. CubeSats are a “standard micro-satellite” (about 10×10cm). Jon got involved with a Kickstarter campaign called ArduSat and helped them design an Arduino board to go in the CubeSat, as well as a prototyping board that students could use in their classroom to design their experiment.
Jon shows the web-based interface where a class would submit their Arduino design, which includes an “upload to satellite” button. Really. :)
(My title for this talk is a little bit of an over-sell, as at the moment the ArduSats have massive waiting queues for anyone to book time on them. But Jon’s intention is to get lots more of them up there, so hopefully that will change sometime soon. I’m not sure where the best place to stay in the loop on that is. I would also love to read a blog or something where schools talked about how their experiments went.)
So his talk (youtube, mp4) was wonderful, highly recommended if you want to feel inspired and amazed and hopeful.
The second is a little more down to earth! I was really into the talk on Visual Editor by James Forrester and Roan Kattouw of the Wikimedia Foundation. (Youtube, mp4, slides) Visual Editor is the WYSIWYG-ish/rich text/Google Docs-ish editor that Wikimedia has developed to avoid people having to learn arcane mark-up and help drag MediaWiki’s interface into this decade.
First of all I think this is inherently interesting, that they have solved (or are well on the way to solving) this problem, as it is one that loomed so large that it has threatened to topple Wikipedia and the other Wikimedia projects many times. Large vibrant communities have certainly been destroyed (or destroyed themselves) over less.
Secondly, they needed to build something that would largely work for the existing corpus that is millions and millions of Wikipedia articles, in over 300 languages. OVER 300 LANGUAGES.
What was most fascinating was hearing about how they tried to reproduce “expected behaviour”. For example, in a right-to-left script, when you press the right arrow, should you move right in the text or “forward” in the text? And in bidirectional text such as numbers embedded in Hebrew, which direction is “forward” anyway? And do Windows apps behave in a similar way to OSX? Or phone OSes? What about IME software for inputing non-alphabetic scripts?
You would think the simple design of moving a cursor around some text was a well-solved problem, but not necessarily. It must have been a nightmare. A fascinating nightmare.
Oh, and if you work on some web-based software and think a nice HTML editor could be an improvement over whatever your default is, they would love to help you try and integrate VisualEditor into another project, like Drupal or Wordpress.