Hardy Heron Alpha 4 Released

LarryBoy writes "Ubuntu 8.04 (Hardy Heron) alpha 4 was released Friday and Ars Technica has a look at what's new in the latest builds of Hardy Heron. 'Although many of the significant architectural features like PulseAudio and GIO are still in transitional stages and aren't fully functional yet, Ubuntu 8.04 alpha 4 is still very impressive. I'm a big fan of D-Bus and I'm very pleased to see it being adopted throughout the entire desktop stack in core components.'"

Comments

Congratulations!

[BeeBeard]

Big congratulations goes to the Ubuntu team for sticking to their release schedules, and getting their fairly solid alphas and betas out there for users to bang on well in advance. Like many others, I thought that Ubuntu Linux was just another flavor-of-the-month distribution, but the tenacity, reliability, and graciousness of the Ubuntu community has proved us all wrong.

-A Longtime Gentoo User

Re: ndiswrapper

[jd]

POSIX simply doesn't support all the facilities required for a trusted (as in "Trusted IRIX") Operating System and, let's face it, security has for a long time been moving more towards MAC and not just for Linux. POSIX got into updating security surprisingly late in the game, and although there are patches for POSIX ACLs for Linux, I can't think of a single distro or mega-patch that includes POSIX security. I seem to recall Linux is deprecating some POSIX functionality, like POSIX ttys, and the kernel has always included unPOSIX-ish code where it is clear that the official POSIX syntax or semantics are carp (sic) or where support for other standards has been useful or expedient.

D-Bus may not be the answer to everything, individual technologies rarely are, and it's not as if D-Bus was even the only user-level software bus commonly used in Linux, but it has interesting potential. Not sure how well it currently plays with clustering technology like MOSIX, or grid technology, but given the effort being poured into developing user-space software buses precisely for those, I imagine that's just a matter of time.

Personally, I'd rather have more localized limited-purpose buses in any case where a general-purpose solution is slower and/or heavier. The code can't be that maintenance-intensive and too much abstraction must eventually pessimize the resulting code. Moore's Law is worthless if code gets slower at the same rate systems get faster. Nonetheless, any general-purpose abstract IPC that is easier to implement against than traditional mechanisms (RPC, CORBA, Unix sockets, System V messages, etc) must surely be beneficial - even if those end up being the mechanisms used under the hood. In fact, the more of those implemented and the better you could switch data between them, the more portable such a software bus becomes as well as the more optimal - to a point. The whole trend in programming is towards such pluggable solutions, it's surprising IPC is so far behind almost every other mechanism out there, and unless there are specific technological reasons to not use a given generic mechanisms (such as performance costs), you're already using so many that are not following some standard or other that it's absurd to discriminate against one just because it's not specifically POSIX.