Disqus - Latest Comments for markbirbeck

Re: Getting Control Of Your .dockerignore Files

Mark Birbeck — Tue, 01 Oct 2019 18:13:02 -0000

I wouldn't put a .env file into an image or running container. How exactly I get keys into a container depends on the environment. If I'm testing locally then I create a Docker Compose file with the names of the keys I want, and then I have a .env file in the same directory. Docker Compose has a great feature that it automatically picks up a local .env file if one is present. If I'm deploying to a Docker Swarm I would use Secrets. I'm not that familiar with Kubernetes but they'll have something similar. And if you're using AWS then Secrets Manager is worth a look, too. For testing, then most services like GitLab/CircleCI/Jenkins/Wercker etc. will have a way to put the keys in some environment variables configuration section that will get passed on to the tests. And when it comes to deployment with services like Netlify and GitLab, you'll have the same thing, and the keys can get passed through to your live deployment.

Re: Probably Wrong | Avoid Else, Return Early

Mark Birbeck — Fri, 30 Mar 2018 10:21:30 -0000

I definitely agree with this. I've also been using this approach for many years but I got to it via a slightly different route.

I used to be a C programmer, and when I moved to JavaScript some years ago, I simply wrote my JS in the same way that I used to write my C. My coding style didn't really start to evolve until I read the excellent 'The Art of Agile Development', which is mainly about Extreme Programming.

There are of course many useful notions in the book and in XP, but one that is relevant here is the idea of 'fail fast'. It's usually regarding as applying to projects and experimentation (probably because of the whole 'lean startup' philosophy), but in XP it's used up and down the chain of abstraction. So I started to apply the idea of 'fail fast' not only to my projects and infrastructure, but all the way down to the level of functions. And of course if you do that you end up with exactly the same layout as you show in your examples.

It was later that I also started to appreciate the other benefits of this approach along the lines that you describe, such as fewer `else` clauses and much better indentation.

Re: Why You Need an Ontology (Schema.org, JSON-LD and ElasticSearch - Part 1)

Mark Birbeck — Tue, 06 Mar 2018 18:13:23 -0000

Mmm...good point Daniel. I shall try to sort that out! Thanks for the prod.

Re: A Mixin Approach to Material Design Lite Using Sass

Mark Birbeck — Thu, 08 Sep 2016 19:00:20 -0000

Sorry for such a long delay in getting back to you!

I've just managed to get some time to revisit the templates. I've fixed the error that you were getting, which didn't show up for me all those months ago, but did now. I can only think that it's because I'm using a newer version of `node-sass` than I was before.

Anyway, if you're interested in playing with any of this stuff the source for this blog is at https://github.com/markbirb.... It has a dependency on Markheim, my SSG, which although very configurable, in its default mode uses a Jekyll layout. The blog also has a dependency on the MDL templates that we're talking about here -- with the fixes to the `@extend`s.

So in theory, simply getting the blog repo and then doing `npm install`, followed by `markheim build`, should be enough to recreate this blog!

Re: Leonard Cohen: age cannot wither him

Mark Birbeck — Thu, 19 Sep 2013 08:51:44 -0000

Great review Mick...absolutely spot on in all respects.

For the record though -- speaking as someone who learned to strum a guitar by playing along with LC in their teens -- his music, poetry and novels were always life-affirming, and particularly about passion.

Oh...one more thing...can we please add Bowie to your Stones and Stewart list? That was then, and this is now.

Re: Choosing a Blogging Platform - Mark Birbeck's Blog

Mark Birbeck — Tue, 26 Feb 2013 17:23:47 -0000

Looks interesting, as does ruhoh.com mentioned before.

Re: Using Knife to launch EC2 instances without a Chef Server - Mark Birbeck's Blog

Mark Birbeck — Tue, 06 Nov 2012 06:14:48 -0000

Hi Timur. `knife-solo` looks very useful. Thanks for the reference!

Re: Using Knife to launch EC2 instances without a Chef Server - Mark Birbeck's Blog

Mark Birbeck — Tue, 06 Nov 2012 06:08:17 -0000

Hi Michael. There are two dots in the EC2 hostname -- is that just a typo when you created this comment?

Re: Using Knife to launch EC2 instances without a Chef Server - Mark Birbeck's Blog

Mark Birbeck — Sat, 19 May 2012 04:33:05 -0000

Hi Brian,

That looks like exactly what I needed! I shall investigate...

Thanks for taking the trouble to point this out.

Mark

Re: Using Knife to launch EC2 instances without a Chef Server - Mark Birbeck's Blog

Mark Birbeck — Fri, 23 Mar 2012 11:56:54 -0000

Yes...I agree.

The quickest fix I could find was to simply disable the entire call to `bootstrap_for_node()`, based on adding a setting to `knife.rb`. But it does mean that I have to use LittleChef to actually deploy Chef to the new server, which is daft since all we want to do with Knife is stop it registering the node on a Chef Server -- we don't want to stop it deploying Chef. So I do plan to try to patch at a lower level in Knife, but that involves going further into the code than I have.

If you get to it before me, please let me know! :)

Re: BBC's Semantic Music Project

Mark Birbeck — Thu, 22 Jan 2009 15:50:12 -0000

We get some (already existing) metadata about which albums a band has produced, chuck in a bit of info about which Radio 1 DJ has played that band the most, and now the semantic web begins?

The semantic web is admittedly a slow burner :) but that's just the way it is; there isn't going to be any one single thing that allows you to say, "ah...that was the day that a real semantic web finally took root".

Don't get me wrong, I like what the BBC has done here, but there's a lot of far more interesting stuff around.

Regards,

Mark

Mark Birbeck

http://webBackplane.com/mark-birbeck

Re: The BBC, microformats, RDFa and Resig

Mark Birbeck — Tue, 24 Jun 2008 11:28:24 -0000

@data-* is not intended to provide 'global' data, it's specifically about giving authors a convenient place to put their own 'local' data.

In other words, the idea is *not* to come up with @data-surname="Dumbhill", and then for everyone to agree that this is equivalent to a FOAF surname (or whatever). It's much more about you using @data-borders="true" to indicate that you want the DISQUS JavaScript to put borders on all of your threaded discussions, and me using @data-borders="portugal france" to indicate to my mapping JavaScript that Span has borders with Portugal and France.

In other words, you can use it for whatever you like, and so can I, because the data is for consumption within the web-page itself.

RDFa on the other hand is all about globally recognised data, and as such is trying to solve a very different problem to the @data-* one. As you can see, they can easily co-exist, since they deal with different value-spaces.

Re: The BBC, microformats, RDFa and Resig

Mark Birbeck — Tue, 24 Jun 2008 11:08:47 -0000

John,

First, RDFa already has parsers that work with HTML...it's not exactly difficult to traverse a tree and pick up some attribute values. :)

Second, this whole thing is something that I've been working to solve for a few years now. There is a very simple way through this, and it's in the CURIE specification, which handles the URI-abbreviation part of RDFa; essentially the CURIEs spec allows for URIs to be abbreviated like this:

a:b

or like this:

That's how we get:

@rel="license"

to work in RDFa. But it means that some language could come along and add CURIEs like 'hcalendar', which could then be used like this:

I won't pretend that this is all a done deal, but it's no accident that there are 'hooks' already present in RDFa and CURIEs to provide a convergence with Microformats.

Some other discussion on this topic is in my blog:

<http: internet-apps.blogspot.com="" 2008="" 06="" microformats-and-rdfa-are-not-as-far.html="">

Regards,

Mark

Re: Yahoo's First Search Gallery Apps Leaves You Searching For The Right Results

Mark Birbeck — Fri, 06 Jun 2008 19:34:36 -0000

Wow...is Erik not a popular guy then? :)

Anyway, I think some of the comments that say that Erik misses the point, are themselves missing the point.

Erik pointed out--and Amit concurred--that if you allow a particular plug-in into your world, then that is a pretty clear statement of your preferences. The argument about spam doesn't stand up, since if I chose the LinkedIn extensions, then I obviously want them, and if I don't want them, I can turn them off again.

Of course that doesn't mean that this is there today, but as Amit said, Yahoo! will be looking into it.

(It's no different to putting the weather forecast as the first result, if I search for "London, UK"; why not put LinkedIn above any other occurrences of a person's name, if I chose LinkedIn, or last.fm at the top if I searched for a band, if I have enabled the last.fm plug-in?)

But the reason I'm responding to this point, rather than just letting it fly by, is that there is an important underlying principle here, which is that semantic information is not just 'tagged data', but also user behaviour. For example, if in GMail I say "always allow images for this email address", I've given a pretty good steer about the relationship I have with that email address. That information is legitimate 'semantic' information, just as much as any structured data is.

The same goes here; it's true that Erik is asking for SearchMonkey to do slightly more than it currently does. But that extra functionality he suggests is exactly the kind of thing that is needed if we are to make the semantic aspect of the web usable.

Anyway...interesting post...interesting service...interesting times...but now I better get back to building tools for the semantic web.

Mark

webBackplane
http://webBackplane.com/mar...

Re: Making the Web Searchable: The Story of SearchMonkey

Mark Birbeck — Fri, 06 Jun 2008 19:06:38 -0000

Hi Alex,

Great article!

A couple of comments, if I may...

First, those writing comments along the lines that it's not yet certain whether SearchMonkey will succeed have a point, I guess, insofar as nothing is certain. But I do happen to think that the guys at Yahoo! are on to something. And even if it doesn't fly, the broader point is that the relationship between data, search and browsing is changing for good now, and there is no going back.

This is an area I've been interested in for a number of years, and is in fact the reason I devised RDFa in the first place. But in the model I've been working to create, embedded metadata is an enabler for users to customise parts of any web page, not just search results.

(If anyone is interested in this topic, I gave a tech talk at Google a few weeks ago, which is available on YouTube at:

http://www.youtube.com/watch?v=mxE3FeOyS-E

The talk first introduces the RDFa syntax, and then shows some demos of the way that RDFa can be used to create richer UIs.)

My second point is just a little bit on the history of RDFa. You mentioned dates for Tantek and Ian, but almost imply that RDFa appeared from nowhere at the W3C!

The first public draft of the syntax (then called RDF/XHTML) was in February 2004:

http://www.w3.org/MarkUp/2004/02/xhtml-rdf.html

although there had been a number of earlier drafts in 2003, that weren't made public. Also, my first public presentation on the topic was later that year, at XML Europe:

http://webbackplane.com/node/57

although once again, I had presented on the subject in private meetings. One example is at the W3C Technical Plenary earlier that year, as reported on xml.com:

http://www.xml.com/pub/a/2004/03/03/deviant.html

Anyway...the point is that RDFa has a solid history, which is at least as long as the other solutions that have been proposed.

One final point; you say in the comment that you've added, that RDFa is complicated, counterposing that people can "actually understand" your syntax. I have to take issue with you there, and point out that whilst RDFa has sufficient power to support all of RDF, it is just as much at home expressing simple structures. For example, the well-known Microformat rel-license is also perfectly acceptable RDFa, and you can't get much simpler than that.

Don't get me wrong, it's great to see this topic getting some air-time, and I enjoyed your article. But as RDFa is rapidly attracting interest, I thought it worth ensuring that your readers had no cause for misunderstanding.

All the best,

Mark

webBackplane

http://webBackplane.com/mark-birbeck

Re: Mark Birbeck is four years ahead of Adobe AIR

Mark Birbeck — Sat, 31 May 2008 09:40:00 -0000

Hi Steven,

Actually the moral isn't at all 'don't trust anyone', and like many other entrepreneurs with a small business, I have no illusions about the protection (or lack of) afforded by an NDA.

The point that myself and Ernst-Jan were discussing (over a very nice dinner :)) was that, although he is disappointed in the lack of vision in start-ups, I thought it worth bearing in mind that you also can't be too far ahead. I explained that our Sidewinder Viewer has been taking shape for a number of years now, and very explicitly sets out to use web languages to build desktop applications--yet when we first started trying to talk to people about it, they just didn't 'get it'. And to emphasise the point, I recounted how even Mike Arrington didn't get it. :)

Yet today we see everywhere blogs that begin "As I predicted a month ago, the desktop and the web are merging". In other words, at some point your idea goes from being far ahead to being mainstream.

But I wasn't moaning about this, and Ernst-Jan captured my view very well; I was saying that the good thing about products like Adobe Air entering the fray is that it helps people understand what we're doing, and hopefully we can use that to further our software. (It's rare for one piece of software to completely win out over all others; even if Air does well, there is space for others.)

For people who are familiar with the ideas of Crossing the Chasm, none of this will be new. It's a fact of life that the best way to promote what you are doing is by reference to something that people are already familiar with, and it's exciting for us that we can now more easily explain Sidewinder, by saying that it is 'like Adobe Air, but open source, and using standard languages'. (They use Flash, we use XForms and XHTML.)

And as Ernst-Jan described, our next phase is probably to move on to saying that Sidewinder is 'like Adobe Air, but open source, using standard languages...and with semantics'. :)

I think the standards angle is a good one for us. For a few years now we've also been doing browser extensions that can be both Ajax libraries and C++ components, and in the last day or so Yahoo! launched pretty much the same thing, with BrowserPlus. I'd claim that again we were ahead of the game, but once again I'm not crying about it, because it will again help people to understand the ideas behind what we are doing; now we can say that our software is 'like BrowserPlus, but open source, and using standards'.

So if there is any moral, it's probably that if you're going to pursue the 'vision thing', be prepared for the long haul, and constantly look for ways to help people understand what it is that you are doing.

Oh yes...and read Crossing the Chasm.

Best regards,

Mark

http://webBackplane.com/mar...

Re: Yahoo+SearchMonkey+Launches+Into+Private%26nbsp%3BBeta

Mark Birbeck — Thu, 24 Apr 2008 15:21:10 -0000

Hi Mark,

Good post.

A friend of mine told me that this wasn't the kind of thing that TC was into, but it seems like you've taken a pretty good look at it, so he may be wrong. ;)

Anyway, I'm doing tech talks about RDFa at Google and Yahoo! in SJ, week beginning May 19th, so if this whole thing isn't old news for you guys on the cutting-edge, by then...well...I'll be in town.

All the best,

Mark

Re: Semantic Web or: why Yahoo! resisted Microsoft’s takeover

Mark Birbeck — Fri, 14 Mar 2008 14:11:18 -0000

Hi David,

In my view, you can't hope for a viable semantic web without being able to publish metadata as easily as you can publish a blog, and although it's taken too many years to get there, with RDFa and Microformats we are finally there.

But you also can't hope for a viable semantic web until you can search for that metadata. Unfortunately, without RDFa and Microformats you're left Indexing RDF/XML files which the the Googles and Yahoos were never likely to do, as long as the return was so difficult to see. It's also difficult to see someone else doing it in a viable way, as long as the amount of content in those RDF/XML files remains a sub-set of the web universe.

However, put these two together, and you have an exciting proposition. Indexing HTML files is what the search engines are good at, because that's their job. So to dig out any metadata they find is not such a big step for them.

And more than that, since the publishing model for metadata is increasingly going to be the same as for publishing any other web page, there will be a lot more data available for the search engines to index, making it a lot more worthwhile to do the indexing.

As you say...isn't it great when a plan comes together. ;)

Regards,

Mark

Re: Freebase Will Prove Addictive

Mark Birbeck — Wed, 04 Apr 2007 21:22:26 -0000

Tim,

Ok...I see what you mean now. In that sense Freebase may or may not be the missing link, but whether it is or not, I certainly agree with you that there's a ton of metadata out there that we could 'infer' and get for free, if we put our minds to it. After all, a 5 year old child can tell that the people who visit our house for dinner should be in the 'friends' section of your address book. They could even work out that the people whose profiles appear on the same page as ours in the departmental intranet are probably colleagues...and so on.

I've been thinking about this kind of thing a lot over the last few years, and part of that work has been channelled into RDFa, which I mentioned before; this allows you to place metadata into XHTML pages, which means that we can start to do some of the things we're discussing here. For example, RDFa means that it's easy for a blogger to add small bits of information about themselves at exactly the point where they referring to it anyway--in some article or in their profile. The tools are not there yet to make it completely easy, but the architecture is.

But another space I've been looking at concerns the 'emergent' or 'inferred' aspect that you are describing, and in particular I've been looking at the seemingly narrow space of the information we provide when we interact with software. For example, in GMail I can click an option to get GMail to always show images in any email that I receive from O'Reilly mailing-lists. Given what you have said so far, I'm sure you'll agree with me that this seemingly simple act--one mouse-click for most of us--tells us a great deal about the relationship between me and the person or company sending the email. (As does the even simpler act of not clicking the link.)

But the frustrating problem is that you can't get to that metadata. The only place that 'knows' that I trust O'Reilly mailing-lists is GMail, and it does nothing with it. (At least nothing that we know of. :)) At the moment it is not possible to make use of this information in some other piece of software, for some other use. As it happens, my company is working on a software platform that aims to allow us to 'capture' those kinds of pieces of information, storing them so that other 'applications' can make use of them, but as you can imagine it's a lot of work.

Anyway, I do now see what you are getting at in relation to Freebase and in your follow-up, and although I can't help leaping to defend the 'big picture' :) I certainly acknowledge that a crucial part of building a (the?) semantic web is to be able to harness the enormous number of 'facts' that emerge organically, as much as it is to harness the 'facts' from enormous data stores.

Regards,

Mark

Mark Birbeck, formsPlayer

http://www.formsPlayer.com | http://internet-apps.blogspot.com

standards. innovation.

Re: Freebase Will Prove Addictive

Mark Birbeck — Wed, 04 Apr 2007 09:58:40 -0000

I have to agree with Jim and others; the Semantic Web is already by definition distributed and sprawling--a key tenet is that "anyone can say anything about anything".

Let's say for example that the pharmaceutical industry decides to create a complex ontology to define chemicals; there's no point in counterposing that to 'folksonomies', and saying that it's further proof of a 'high-brow' semantic web disconnected from the rest of us. The point is we need both evolving ontologies and we need ones that are created by experts in a field.

And as well as the data, we also need the 'glue' that makes this data available in different ways. I believe one of the most important developments that is being worked on at the W3C is RDFa, a way of using a small number of attributes to add simple or complex metadata to XHTML documents. RDFa is being worked on jointly by people from the HTML and RDF communities. (See my An Introduction to RDFa, Bob DuCharme's Introducing RDFa and the RDFa Primer, for background.)

No disrespect to Tim but I think what often happens is that people look for the one single thing that is somehow going to change the landscape, when the reality is that technology develops as a combination of both evolutionary and revolutionary steps. The same idea may be revisited many times over the years before it finally becomes a reality. It's as true in mathematics, biology, physics, engineering, etc., as it is in software development.

So Freebase is great, and a nice idea. But it's certainly not going to 'bring about' the semantic web. For that we need a combination of many different things (all of which do seem to be progressing together quite nicely, though).

Regards,

Mark

Mark Birbeck, formsPlayer

http://www.formsPlayer.com | http://internet-apps.blogspot.com

standards. innovation.

Re: Open source developer warns about Adobe…

Mark Birbeck — Sat, 03 Mar 2007 19:37:15 -0000

In my view there are actually two issues of interest here, and I would like to suggest that the wrong one is being talked (and worried) about. :) There is indeed a question of 'open source', but the more important issue--in my opinion--is that of open standards.

My company has also been working on a desktop application framework--called Sidewinder (screenshots/a>)--and have been doing so for a few years now. To create standalone desktop apps, authors can use either XHTML or JavaScript (the latter being scripts that simply stand on their own). But the approach we've taken to giving programmers more power over the OS is to use standards as much as possible.

For example, to create a system tray message we use the XForms 'message' action handler; to save a document to the local file system we use XForms submission; to communicate between different 'internet applications' we use DOM 2 Events; to get speech we use XForms messages with CSS. And so on.

The idea is that in the same way that web pages can invariably run within any browser that conforms to open standards, so too we can create 'internet applications' that will run in any 'desktop framework' if they too are built using open standards.

In a sense we're turning XHTML into a programming language, and as a working name for this, we've called it 'xH'--a profile of XHTML that includes XForms, SVG, RDFa and XBL.

This language profile comprises a set of XML schemas, built using techniques defined in XHTML Modularisation 1.1. This means that 'programs' written with xH can be validated in editors such as oXygen. Although all of the pieces aren't yet in place, I think that so far at least, the whole concept stands up quite well.

Which is why I believe that open sourcing Apollo is of less significance than what languages Apollo uses, and in particular how it makes the OS 'available' to the programmer; ideally the way to 'open a new window' will eventually be the same in Apollo, XUL Runner, Sidewinder...and any other desktop application framework that will almost certainly come along.

Anyone interested in reading more on this approach might find the following of interest:

A Standards-based Virtual Machine, a paper I presented at the W3C's Web Apps workshop, three years ago;
xH: The new programming language you already know on the formsPlayer site;
Is it AJAX, Ajax...or XForms?, on my blog.

Regards,

Mark

Mark Birbeck, formsPlayer.com
http://www.formsPlayer.com | http://internet-apps.blogsp...

Re: Explaining the Semantic Web (Or Die Trying)

Mark Birbeck — Mon, 26 Feb 2007 17:56:57 -0000

I just love this 'live' style of writing! Thanks for the mention, which I'm taking as almost positive, whether you like it or not.

Mark

Re: London Mashup: What's Next, Web 3.0?

Mark Birbeck — Sun, 25 Feb 2007 19:33:23 -0000

David,

Thanks for the mention in the write-up.

I hope you don't mind if I link to a post of mine that has a screenshot of the demo I was doing. And since it sounds like some of the things I was saying weren't clear on the night, there are also some notes on the whole "why would I want to turn a web application into a desktop one" issue that has come up in a few write-ups, and they are over in the comments section of Jag's photo of me. (Don't worry, the discussion isn't a discussion of the photo :) but relates to a comment that Jag added alongside the photo.)

Thanks again!

Mark

Re: Plugging In to the Documentation

Mark Birbeck — Sun, 04 Feb 2007 21:29:00 -0000

I just created a toolbar for IE that allows you to go directly to a page in the Prototype documentation using a ‘search term’. It was built using XForms, and a tutorial on how it was done is here.

The toolbar does everything you need with only one XForms control and one action handler, which means that there’s plenty of scope for adding additional features like a list of previous searches, and so on. I’ll also be adding a few other ‘applications’ that make use of the Atom data.

I know that most people will be smugly using Firefox with Firebug. :) But if you do use IE…well, now you don’t have to be so envious, since you have a toolbar available if you want it.

All the best,

Mark

PS I should add…great site!

Mark Birbeck
CEO
x-port.net Ltd.

Re: Webified Desktop Apps vs Browser-based Apps

Mark Birbeck — Sat, 09 Sep 2006 10:54:48 -0000

The fact that this topic is getting discussed is great. I think from the comments and views that I've read, that one permutation is missing, and that is the creation of an additional layer above the desktop that web applications can make use of to turn themselves into desktop applications. A key part of this idea is that this layer should be standard.

That's what we've done with Sidewinder; we've created a cross-platform environment that supports features such as creating windows that dock to the side of the display, can auto-hide, have transparency and opacity, and so on. Sidewinder also allows events that are fired in one window to be registered for in another (allowing applications to be built from smaller pieces).

However, unlike traditional desktop applications that need to be written in C++ or Java, Sidewinder applications are written in either XHTML or JavaScript...or more often than not, both.

So a web application could be as simple as an XHTML file that contains a page with a clock in it, through to a complex application in JavaScript, that in turn creates many XHTML windows and manages events passing between them.

We call these applications 'internet-facing', since they want to be desktop applications, but they have so many internet-related features that they are easiest to built with web technologies such as HTTP servers and XHTML pages. By providing a framework that takes this into account we can get the best of both worlds.

I mentioned that a key idea was the use of standards, and Sidewinder uses XHTML, XForms, SVG and MathML to build applications.

If anyone is interested, we have various demos that show how easy it is to make a web application into a desktop one:

<ul>

<li>KoolIM as a desktop chat application;</li>

<li>Technorati Mini as a desktop application;</li>

<li>Enhancing your Google Calendars.</li>

</ul>

Best regards,

Mark

Mark Birbeck

CEO

x-port

http://skimstone.x-port.net/