I just met Copy and Paste Math|MathJax which is very interesting but really:

it shouldn’t stay so!

Copy-and-paste MathML should happen as soon as you invoke the copy command after having selected (part of) the formula.

Hopefully, MathJax will consider implementing that part as well.

In this video, I find the paste phase pathetic: one really needs correct media-designation in the clipboard so that such is only needed when you have special desires. That one is a normal desire.

Quite a few polishes in: - Includes input-editor enhancements; Drag-and-drop of formulæ now works again. - Dutch and French internationalization polishes - Groups should be mapped from Moodle courses.

And a lot more, see Jira issue list towards 1.1.

Revision 36 should come to all authors.

A fascinating discussion has been started by Tim Berners-Lee on the TAG mailing list, here’s the pointer to the archive: http://www.w3.org/mid/AFFAB130-B693-4AC9-91E6-B6834E57B3F5@w3.org

it is based on a blog post by John Gruber, Tynt, the Copy/Paste jerks which is also fascinating as a rebellion.

I observed the following fancy bit on a wolfram alpha’s answer:

Current temperature in Lyon: -6 x 10^-6 °C (degrees Celsius)

Tastes like monstruously freezing ;-).

See picture.

An introduction video of jEditOQMath used to author ActiveMath content can be seen here:

http://www.activemath.org/~paul/tmp/jEditOQMath-overview-web.mov

(379 Mb, MOV file, plays in Quicktime Player and VLC).

Revision 11 is posted on ActiveMath-SVN.

It now displays roots in HTML (but fails at nesting of roots). It also attempts at displaying the port number in amStartup on unixes. Finally, the notation elements now also work in MathML.

To update… svn update as in ActiveMath-SVN. Nicer display of roots may need an index rebuild.

I’ve just brought the trunk of ActiveMath, as can be seen on devdemo, on the ActiveMath-SVN-stable. This brings a large amount of fixes and features:

• polished MBase reload, most of the issues that were reported are solved
• content-collection-based menus when you are an author
• hidden user-model per default (no more confusing bullets, can be changed in view options)
• enhanced cache invalidation: you can now reload and request the objectives selection menu again
• bookmarkable course-generation requests (can be given to learners for example)

• new StaticBookMaker makes the collection-info, e.g. the list of files
• content-collections now have a collection-metadata (which is inherited) by default
• newly created content-collections now have a creatie-commons-attributions-sharelike by default
• templates are now organized in three sections: documents, items, formulæ (thanks Yecheng)

The need is extreeeeemely common: you want a map of some region to be included in the next flyer, in your web page, in your learning material. The map should be of good quality, be realistic, and certainly must contain your own annotations. How can you do?

Until yesterday I only knew the answers such as the Google-Maps-Embeddings: very limited annotations, extremely constrained licenses (no flyer for example), pretty slow, … but now I discovered OpenStreetMap.

I’ve been musing with ImageMagick and found a little snippet to produce drop shadows on the command-line with this amazing open-source library.

Given my file input.png I can obtain a version with drop shadow as target.png by running:

convert -depth 8 -threshold 0 -negate input.png  -bordercolor 'rgba(255,255,255,255)' \
   -border 20x20 -gaussian 0x7 -shave 10x10 - | \
 composite -geometry +5+5 -gravity northwest input.png - target.png

A nifty feature has just emerged on daily activemath builds and jEditOQMath: advanced metadata inheritance. Basically, it allows an author to avoid most of the repetition of metadata by simply putting that metadata in an enclosing element.

This is the story of a confusion which grew from each other’s copying a “nice example”: it’s about what is the normal notation for the binomial coefficients in Russian and French.

It starts as a paper we are writing, which copied another paper of mine (at MathUI), itself copying others sources, without precise citations (1, 2, and 3), the most authoritative one being MathML 2: all these sources claim that in Russian, the binomial coefficient (the number of subsets of cardinality k within a set of cardinality n) is written while in French, it’s written .

I came accross Why PHP should never be taught which is hilarious!

I always had a small mistrust of PHP as being a too easy language but this post is a perfect illustration of how things can go wrong if you have the slightest ignorance about all the details of the environment you are programming for! As a summary it says… equality-test in PHP is not transitive!

I am in love with the rsync command-line tool. It’s a careful tool, a brilliant implementation, and well-thought command-line interface that surpasses cp, scp, and many others.

This is a very nice experience I am having: I am editing writing an article in LaTeX and that is with OmniOutliner, the incredibly nice outliner of the Omni Group. Luxury is there to manage the texts, to hide parts temporarily, to apply styles and to navigate while the quality of the LaTeX output remains.

Recently I encountered two websites which are really for educational purposes, even for math education purposes, two educational entreprises that aim at making mathematical content web-accessible. But it really is only “user-browser-web-accessible”… two examples of the wrong approach of rich internet applications.

David Carlisle informed us today that:

chapter 6 is now chapter 7…

(indicating that we now have reorganized the chapter 6 of MathML specification to become the chapter 7 and vice-versa).

This had the topic “end of an era”.

Ridiculous?

Not at all. Funny it is but the fact that it needs a mail with such an emphasis proves the care taken to develop the MathML standard at W3C: the wish of ongoing compatibility goes as far as chapter numbers. And indeed, it really means that people speaking of chapter 6 will have to change their words.

Precious fun!

The Semantic Web is not Restricted to Ontologies!

The semantic web is a nice strand. Something where AI should be able to do a huge job based on the massive knowledge spread around the web. Is it about AI or more generally machine processabiity? The last call for paper for the WWW conference actually broadened the semantic web track call to “Semantic / Data Web”.

Over and over the discussion of JSON vs XML (vs YAML…) appears: they are compared as languages to serialize object graphs. I may be old fashioned but I find there’s no valid reason for anything but XML.

Here’s a set claims in favor of JSON (or YAML it seems):

• JSON is more readable: I am sorry but I don’t find this true, being XML it’s much easier to know where parts start and end since end-tags are named. Most of the times the object graph is not sufficiently indented to recognize any structure hence readability only appears after you’ve tweaked a bit.

Translation tools are many… but what good are they? To my surprise, there’s always some folks that find a tool that sounds cute but then it never helps. Let me try to identify the hard questions of translations’ management:

• workflow of updates (by developers) and management of translators’ contributions
• recognizing context when translating phrases
• managing translation duties for documents and phrases

The fancy conclusion? No tool I’ve met really helps any except the management of duties and generally they do so by making completely impossible the others.

By chance, I noted that my daughter Pénélope’s math-book had exercises for geometry “with software”. So I gave it a try with her: she manipulates the mouse and tries… I just guide. It worked real well!

You know what?

A content-dictionary as can be found by tons on http://www.openmath.org/cd/ is mostly a social artifact. It’s a set of descriptions of symbols so that one can mean what others means.

It’s not really new that mathematical notation is made there to be abused so as to be most efficient for the current context. But I just met an extreme case:

In the OpenMath3 and MathML3-content efforts, we are polishing the description of symbols’ so that they can be common. Discussion about using linear syntax appears, of course, and an extreme case about the need for context was just posted by James Davenport:

Skype has always appeared to me as the real end-user application… something you have to accept to reach certain users. It’s nicely designed but it’s too proprietary.

But hold on, Skype actually tried a standard… sed!

I just attended the MUPPLE workshop here at ECTEL in Maastricht. In general, I tend to be finding bad style the mashup approaches such as the “very simple web-page embedding”. In this case, I was impressed of seeing something real useful happen just under my eyes.

Since a few weeks, we are playing with Google Web Toolkit, a very nice tool that compiles java code to JavaScript. Among the features of GWT, transparent remote server invocation is offered. The nice thing is that it right away forces you to consider services asynchronous which yields responsive user-interfaces. The wrong thing is that it seems to be fully version dependent to a point where, basically, one wishes to throw it all away.

Sorry, this time it really is unrelated… but I just wanted to celebrate the joy of OmniWeb again and again.

OmniWeb is a commercial browser, not very expensive, one with customer support and a polish that you never get anywhere else for many aspects! It has secret features for web-developers and this fluidity that pushes you to abuse it…

I recently had a very simple request… soooo simple: our user just wishes to copy the formula from Mathematica (which can copy it in MathML) and paste it on something that does web.

I just went around and tried… SeaMonkey should support that in editor and reader: copy a piece of HTML with MathML and paste it, didn’t even work… my 1/x became a place full of nbsps in three lines!

Tim Bray indicates that almost all the legal impediments before an open-source java are cleared. Now this really means there might be, one day, a clean java, something that does cause headaches to distributions!

I much love the GNU Classpath license… really… as simple as that: do whatever you want to “link to it” but there rest is the GPL-full-land.

There’s a wind for more content construction in the ActiveMath group, with at least two projects at the University focussed on creating content (and a adapt platform and…). And MathML starts to play an important role there.