Archive for July, 2007

Links of this day

By way of Daring Fireball: Typographica has an article on their »favorite fonts of 2006« and Mark Simonson has put up a talk he gave at SOTA TypeCon 2006 on Adrian Frutiger.

BoingBoing noted the existence of Toypography blocks, an intriguing collection of reconfiguration glyphs.

While I haven't seen an announcement yet, the beta version of luatex is either out (it has a tag in the Subversion repository) or will be very soon. I was tempted to start using it for my dissertation, but it looks like it is incompatible with some of the things I have been doing to support UTF8, and it would probably waste more time than it will save me to redo that and handle all the other incompatibilities that I will inevitably discover.

From Chris, I learned about the new »glasses-mode« in GNU emacs. Essentially it will insert a custom separator into CamelCase identifiers. It's pretty slick. At some point I should probably look at the source to see how it works, since it would be nice to configure emacs to provide »views« for other sorts of information.


New otftofd

I just finished a new release of otftofd, my tool for generating LaTeX NFSS infrastructure for OpenType fonts (with a particular emphasis on those that have a large number of optical variants). It is available using darcs via

darcs pull

The most significant change in this release is that I have removed the dependency on the Cash library. While useful, just getting OCaml built and installed is already more complexity than most people want to deal with. The second change is that I've added an option --fsuffix= that allows for appending a suffix to the name of the font family. This is particularly useful for allowing multiple versions of the generated files for the same family. For example, in my dissertation I use

otftofd --enc=T1 *.otf
otftofd --enc=T1 --osf --fsuffix=OSF *.otf

so that I can refer to a version of a typeface with lining figures as the default and a version with old-style figures as the default. Finally, I've added an option --scale= that instructs it to use a scaling factor expressed with respect to 1.0


Helvetica at MoMA

Helvetica Bold, 36 point, sorts

Yesterday, Brian and I made a trip up to New York City to see the Helvetica exhibit at MoMA. It was interesting, but the exhibit was even smaller than I expected. Beforehand I tried looking for information on the web about how large it was to gauge how long it would take, but everyone that wrote about it didn't really go into that detail. So, for future reference by others out there, this photo shows the entire exhibit.

The entirety of the Helvetica exhibit

Probably one of the parts of the exhibit that I found most interesting was this diagram they provided for comparing Helvetica with Akzidenz Grotesk.

A comparison of Helvetica and Akzidenz Grotesk

Since I have been asked a few times about software for comparing typefaces, the use of translucent overlays struck me as a useful basis for developing such software. More photographs from the trip are over on Flickr.

Another unfortunate thing is that they are not doing even irregular showings of the Helvetica documentary at the museum. Hopefully, I'll be able to find it somewhere eventually. Apparently the film was shown in Philly, but I missed it.  It looks like I might be able to catch it in DC in September.

Comments (5)

Bundled links

Some people might find this poster of the Unicode Basic Multilingual Plane interesting. Perhaps I'll get one for my office. I seem to recall that I read about this on a ConTeXt mailing list a while ago.

I also ran across this post on Boing Boing about »Giant Graffiti Typography«.

You have probably already heard, but the GNU General Public License, version 3 has been completed.

Also, all the hype lately over the iPhone has had me thinking about what I am going to do about phone service in Lausanne. I'm pretty sure my Treo is locked, and I am frankly getting a bit weary of using the antiquated PalmOS.

However, given the very lame development platform provided by the iPhone, and the fact that it could conceivably not be available in Europe this fall anyway, I'm considering other options. The ability to run third party software is essential, as one of most important applications on my Treo is Keyring. However, there are two very promising options Trolltech's Greenphone and OpenMoko. OpenMoko is actually a platform rather than a device, and Greenphone is mostly just a device to provide an SDK for Trolltech's particular platform. The only OpenMoko supported device on the horizon is the Neo1973, but it seems to have a lot of potential. The first development models will be available in the next few weeks.


InforML Complete

Department espresso machine mortally wounded

It has taken nearly ten months, and numerous changes in direction, but I am finally declaring my implementation of the InforML language as complete. Or at least as complete as I am planning on making it, modulo bug fixing and polishing that arises while I work on the examples I will be including in my dissertation.

Since I last wrote about InforML, which to my surprise has been exactly a month, I've fleshed out remaining details concerning typechecking and implemented a simple compiler to mzscheme. At Stephanie's suggestion, I also took a day or so and implemented some simple local type inference, which takes care of quite a few of the label and type applications that were previously required.

Typechecking actually became problematic as Twelf proved to have very unpredictable behavior on very similar queries. Just changing a single label in a query would cause it to diverge. I eventually even discovered a bug that was causing the theorem prover to claim proofs of clearly unsolvable queries. At least I see no way they could be proved and the Twelf tabled logic programming engine agreed that there were no solutions. So even if I am wrong and they may be proven, either the theorem prover or the tabled logic programming engine cannot both be correct. Right now I am betting on the prover being wrong because doesn't generate proof terms. To help with both of these issues, I wrote a number of custom constraint and subsumption solvers in SML that handle the vastly more common case where the problems are entirely constrained to ground labels, kinds, types, and signatures.

So now I am beginning to mostly focus on the actual writing of my dissertation along with studying idioms and methodologies for programming in InforML. My current timeline calls for this phase to be completed by the beginning of September, which seems believable.

Comments (5)

« Previous Page « Previous Page Next entries »