It’s been a quiet couple of weeks on the software installation and testing front (apart from NetworkManager), mostly because I’ve been on the road four out of the last five weeks. But having had all of last week in the office, I began to get antsy and have been considering and/or planning a couple of new installations. In case you’re curious (and you have to be pretty nerdy to be interested , they are:
- GNU/Solaris / Nexenta:
The context to this one is interesting. In the interests of being well informed and covering all the bases, I try to run as many operating systems as possible. While the exact operating system topology varies as I wipe and reimage test machines, at the moment I have machines running Windows 2000 & XP (no Server 2003, but am planning on Vista ), Gentoo Linux, Fedora Core, Ubuntu, and Solaris 10. At other times, I’ve run both Red Hat and SuSE, along with maybe three or four other Linux distributions.
The Solaris 10 box is, as you may have read, a V20Z that we have colocated down at a datacenter in the Denver Tech Center. This box has, to date, been rather underutilized – I’ve played around w/ Apache, Instiki, MySQL and Ruby (via Blastwave) and it’s generally been good for me to get used to the similar but quite distinct Solaris 10 environment.
When Cote and I were trying to build a project that we’ve been granted access to during his visit this past week, however, I found myself strangely reluctant to use the Solaris 10 box – the most powerful one we have available, by far – for the task. Why? Because it’s still a bit alien to me. The package management system offered by Blastwave is adequate, but installs its applications in non-standard areas, and thus I’m reluctant to use it extensively.
Enter Nexenta. For those of you that haven’t tracked this project – which is a lot of folks, near as I can tell – it’s essentially a blending of the GNU toolset, the Debian package management application, and the OpenSolaris operating system. From my perspective, this is ideal: I get to continue running OpenSolaris, but under a package management and tools system I’m comfortable with. While Catherine doesn’t appear to agree with me, I find the package management system and its functionality critically important in how I run and maintain a given operating system. So important, in fact, that I’m willing to schedule an appointment at the Tech Center w/ our colo provider, drive down there, and spend a couple of hours blowing away an existing install of Solaris and laying down Nexenta.
Why? Because it’ll be that much more usable for me, and I’m more likely to use the hardware for existing projects in future. If anybody out there’s got Nexenta experience on a V20Z, or has seen a wiki instance where that’s covered, let me know. I really hope it has SSH enabled by default on install, unlike S10 (at the time I installed it).
- Personal Wiki:
While I strongly agree with Luis that Tomboy’s killer feature is its real time save capability (just start typing and its captured, similar to what OneNote offers), my need for a local wiki instance is growing more and more pronounced. Tomboy’s performance at startup is an issue – it chews up a ton of processor cycles on entrance into GNOME – but it’s rather a lack of a couple of features that make a wiki necessary. It’s not that I don’t admire Tomboy’s simplicity, I do. And I’m certainly not asking for something hugely complex. But tagging is huge for me, and I can’t see how – in the current UI – Tomboy could add that effectively. It’d be nice to have a more persistent and broadly functional UI than what a simple panel applet will allow as well.
So while I agree with Luis, I think it will ultimately be easier for one of the Wiki projects to add Writely style auto-saving than it would for Tomboy to meet my UI needs. Looks like I’ll be running an Apache or LightHTTP instance soon.
Incidentally, if you’re looking for an opportunity to do some innovative coding for profit, for open source, or both – this is one area I feel is definitely underserved.
This is what I was preparing for when I crippled my X install last week by mucking around with the graphics drivers in my kernel, but within a week or two I hope to have a working install of Xgl now that I’m told it supports my graphics chipset.
As I’ve confessed to a couple of people, I am indeed a sucker for eye candy and Xgl’s got that in spades, but would I really be doing several hours of compiles just so that I could turn my desktop into a cube? No. But Xgl’s not just about gee-whiz eye candy; it actually allows for things that make your desktop more productive (the Expose-like functionality) or readable (drop-shadows, transparency, etc).
The only thing giving me pause at this point is IBM’s Mike Dolan’s report that Xgl is a major resource hog on his KDE implementation. This is interesting, b/c it was my understanding that by leveraging a modularized approach to the X server, Xgl was actually supposed to be faster. Will have to do some more checking on this.
When will all of this happen? You already know the answer: whenever I can steal the time
 Incidentally, if any of the Wagged people are reading and can ship me one of the betas of Vista I’d be most appreciative.
 As an aside, it’s very interesting to see the affinity that IBM has for KDE; nearly all of the Linux desktop demos I’ve seen them give are on KDE. It’s a bit curious seeing as the major enterprise Linux desktop vendors (Red Hat, Novell/SuSE and – at one time – Sun) were seen to be most aligned with GNOME, but KDE fans should take note.