feeds | grep links > OH Supremes Pierce Jurisdictional Boundaries, MS Sneakily Installs a Firefox Add-On (Again), Ubuntu for Tablets, LLDB as Fast as GDB, and More

  • Top court says businesses may sue residents of other states in Ohio over Internet comments
    HT Chris Miller. I thought there was a pretty strongly established practice, at least for some types of complaints, of a case being heard where the defendant is located. Or am I confusing that with weapon choice and duelling? At all events, expect cases like these, fighting over who has jurisdiction over an interaction that takes place exclusively online. Is there any good precedent for the defendant in Virginia to contest the ruling by the Ohio Supreme Court?
  • Microsoft sneaks a Firefox extension into an update–again
    As Emil Protalinski at Ars Technica explains, at least this time the notes on the update are targeted at an issue with an extension itself. The problem is that the knowledge base article doesn’t exactly say so in as many words. Worse, if the add-on or extension is not present, the update installs it. It does so without permission which is simply insult on top of the injury of “fixing” software that isn’t event install.
  • Canonical working on an Ubuntu version for tablets
    Via Slashdot. Makes sense and seems consistent with the version targeting netbooks. I am consistently impressed with screenshots of that version, makes me wish I had a netbook to give it a spin. Hopefully the tablet version will be that polished and whizzy.
  • The flip side of Apple’s relationship with Open Source
    Via Glyn Moody on Identi.ca, The H has a piece balance some of the criticisms I leveled about Apple’s poor handling of their (entirely legal) re-use of the Readability project’s code. I’ll give on the competitive pressure front but not so much on the enabling, the much belabored example of KHTML/WebKit. I suspect Google would have still created a browser if WebKit had evolved at Apple’s behest. They might have even adopted and helped improve Mozilla’s components which I would argue would have been better for the state of open web standards.
  • Can privacy, social media and business get along?
  • LLDB, relatively new sub-project of LLVM, already as fast as GDB
    Slashdot has the details and links to the projects’ pages. If the scripting languages being ported and built on top of LLVM can access and benefit the debugging capabilities LLDB brings to the table, then I think the argument goes well beyond Clang/LLVM replacing GCC into LLVM driving the velocity of a lot of language and tool development more broadly.

LLVM Project Releases Version 2.7

I wrote about a project built on LLVM, VMKit, on Monday so was clearly primed for this bit of news, that LLVM itself has pushed out a new release. John Timmer at Ars Technica covers the details. As Timmer notes, 2.7 is a major milestone for the project. Its Clang compiler, which provides support for C, C++ and Objective-C, is far enough along now for LLVM to be built using LLVM via the Clang front end. This is known as bootstrapping and is a key step in the maturation of a compiler.

Timmer also explains Apple’s involvement with LLVM yielding high quality support for Objective-C and for the ARM processor, including that chip architecture’s vector instruction set, NEON. He’s very clear in pointing out that Apple is not the only software vendor supporting and building on LLVM. Unladen Swallow, Google’s attempt at speeding up Python for their numerous projects using that language, is also built on top of LLVM.

I am delighted that LLVM has the potential to run and compile code written for Apple’s OS on other platforms. This is possible because of the shared ancestry of Cocoa and GNUstep. I doubt this helps much with the newer iPhone OS but I’d be curious to see if an LLVM based project evolves that helps make such applications more portable. It would be a nice end run around Apple’s ridiculous policy against code generators and 3rd party interpreters.

There are a lot more details in Timmer’s post as well as in the release notes.

VMKit: JVM and .Net on LLVM

Via Hacker News, VMKit is a demonstration of the LL (Low Level) nature of the LLVM project. Rightly or wrongly, I tend to think of the relationship between certain high level languages and LLVM like the one between certain lower level, compiled languages and GCC. In the effort to generalize and optimize across languages, some interesting features and capabilities arise that might not have with language specific VMs and compilers.

(My characterization, even after just a few minutes perusing the LLVM web site, is not entirely fair as it can also be used for static compilation. This doesn’t affect my point, below.)

VMKit is a research project at present. It is some way from being a drop in replacement for either the JVM or .Net’s CLI. Even with that caveat, the project seems to have made some considerable progress, noting that on the Java side it is quite capable of running some non-trivial projects like Tomcat and Eclipse.

VMKit in a way shows how LLVM can run circles around the JVM and .Net. Both of the latter two have projects for running languages other than the ones for which they were originally created. Looking at the projects based on LLVM and the sort of advanced research that makes use of it, VMKit’s choice of LLVM seems highly appropriate to me.

I tend to think that language choice is much more driven by developer preference than any objective difference. LLVM and VMKit make me wonder if the future of programming might see the objective differences genuinely fade away. I like the idea of optimization being targeted at LLVM well below the concern of popular languages while those same languages all enjoy the mutual benefit. I suspect there are still enough legitimate differences that this idealized common platform may never be completely practical. It would merely be nice to end at least one class of hacker religious wars by pointing out everything ultimately runs off the same low level code.