Search options (Page 1 of 10)

Yes you could run those "make a USB stick" programs, or just run through the regular crunchbang install, selecting an external partition as the install location.
And end up trashing your existing installation, because these programs jerk around with MBRs and so on.

Also, for anything more complex, like installing on GPT partitions, and especially getting anything to play with macbooks, EFI, and so on, you need to do it by hand.

The documentation for all this stuff is either non-existent, or encyclopedic (who has time to read the grub manuals, let alone try and make sense of them), and the internet is worthless - full of forum posts that are almost your config, but never quite, and when there is a match the posts conclude with something incredibly frustrating like "got it working now, thanks for all the fish".  Or worse, a howto is actually wrong - For example, I've just spend a couple of hours getting persistence to work, I based my setup on an article at "pendrivelinux" that said the persitent fs should be in "casper-rw".  And that's just the latest little wrangle.

Anyway, useful info in this post (actually I arrived at live-rw by looking at the scripts/live in initrd), I'll add my own stuff when I've got it working.
Specifically I avoid grub like the plague, and have built a USB drive using only GPT partitions, using the syslinux gpt MBR and setting a legacy BIOS bootable flag on the linux live partition which is then running syslinux of course.
On the same drive I have a Mac OS Lion install/maintenance partition, and another EFI partition that I hope to be able to use to boot crunchbang live on a macbook, without using BIOS compatibility mode (which doesn't support USB) - though that's looking like a stretch at the moment.

Is this objective to cram as much as possible onto a single machine?  Just the existing servers, or desktops too?
I would say VB is maybe not the most robust way to do this, and I would also not use the base system as a webserver too. 

Xen, VMware ESX etc would be more industrial strength I think.

I had an idea to virtualise my laptop, ie have it running some minimal OS (crunchbang, in fact) purely as a host to virtualbox, with all the real stuff done in virtual machines, perhaps one per virtual screen or something like that.

My idea of the benefits was that I could more easily boot these clients on different hosts, make copies, etc, etc.  The experiment sort of worked, but wasn't really robust enough to use in anger - I certainly wouldn't take the same approach with business critical server infrastructure, without very extensive testing.

Just my 2p, but good luck anyway - it's certainly the way of the future.

Another vote here for thinkpads.
The T41 was voted "best laptop ever made" by one of the respectable magazines, and the T42,43, continued the tradition.  All really excellent machines, and the keyboards just can't be beat.
The HP competitors, such as nc6000 are nearly as good too, some are bold enough to claim the HP keyboards are actually slightly better.

This is really an excellent project, and I want one.
HDMI is the obvious choice for display, as almost everything new has it, and composite works for the rest.
I do wonder if they'll really get to $25 though, I suspect it'll be more like $50-100 to buy.

The main problem I see is that this might not be any more suitable for education than any other computer.  The generation who learned to program low-level on 8-bit home micros did so often because they had to, and quite often with an eye to creating a hit application.  Programming "for real" on a modern computer is a lot more complex in many ways, with a myriad of APIs, libraries and so on.  Environments like Scratch, or Squeak are perhaps better for learning.

A better hardware tool for teaching would be arduino, especially now that google have adopted it as the android hardware peripheral standard, which gives it the added spice of being "for real" rather than just a toy teaching tool.

+1 for pytyle
I changed the key bindings to switch between windows to be vi-like, and also added another window layout, "none", which just puts the windows back in their untiled position, but without disabling pytyle (so the key-bindings still work).

.

The new zram module appears as just another block device.  In fact you could put a filesystem on it and mount it as /tmp or whatever.  So yes, conky should see it.

Running 2.6.36-4 liquorix under virtualbox 4.0.4 on a mac book 3,1, OS X 10.6.7, the system hangs frequently, it seems some javascript or flash or perhaps just busy networking under chromium is tweeking something.
No such problem with 2.6.32-31
Shame, as I'd like to play with zram swap.

I think there's now a 2.6.37 version of liquorix, and also the latest debian testing is 2.6.38.  How do I get those, and what are the implications for the rest of the system - I don't want to upgrade all packages to testing versions!

Also, flitting between kernels is a real pain under virtualbox, as you have to re-install guest additions each time (the install removes the modules that got built for the previous kernel, + I don't know what else).
I tar'ed the modules for liquorix before running the install again on 2.6.32, and replaced them after running the script, so if that's the only kernel-dependent thing, guest additions should wokr whichever kernel I boot into.  Will test it next reboot - hopefully not too soon.

Guilty as charged, though I did say I thought it was "conceptually broken", in that there is no clear separation of functions, the dock icon serves several purposes depending on context.

Obviously this is what the majority of people want, or at least this is what vendors want to provide, since all the major OS's have adopted this approach.
The paranoid in me says this is a marketing ploy to keep apps in the forefront of users conscious, and Shuttleworth's comments about doing Unity because it's what the manufacturers want is worrying.

The real problem though is that it looks like gnome3 is far more locked down than previous environments, and you won't be able to easily change the way it works.  Very Apple.

Awebb, I agree about transparency, but what about navigation.
Do you navigate/select what you want by choosing the application, or by choosing a window?  What if there are several windows open for a given application, which should it switch too?  What if there are no windows?
A good example is terminal.  Should I navigate by selectingthe "terminal" icon, and then either having to choose one of the open sessions (or perhaps, as with mac OS, just switching to the "most recent" one or whatever the algorithm is).  Or should I navigate by having a list of my open sessions (even though they are all under a single application), and selecting which session I want.
To me, it's not even about windows, but about my content.  I want to select my content, not worry about what application is running it, although of course I want full transparency about which application is actually behind my content.

Yes, I don't see the benefit of stitching two monitors together into a super-sized screen.  With monitors getting bigger and bigger you need the opposite, if anything, essentially a tiling window manager, but based on desktops.
Having multiple virtual desktops on a single screen would let you easily have side-by-side maximized windows without the app having to support it, eg web browsers, and also be easier to tile some regions and not others, eg with pytyle.  Perhaps some tiling managers provide basically this functionality anyway, but with the added complexity of having tiled-regions and non-tiled regions or something similar; yet another layer, when virtual desktops could be used for this.

A start would be one desktop per screen though, which I think is what you're asking for.