User:Tonyr/Journal/Aug07

02Aug07

Netinstall
The mini.iso file in the netboot file collection is a bootable CD image that will orchestrate an Ubuntu install using a remote repository. The most reasonable (and recommended) repository is the one in the Unitesd States (there are repos all over the world). It may be hat there are closer repositories than the standard one, us.archive.ubuntu.org. There may be one at OSU. There should eventually be a local repo at FreeGeek. That would be nice, but in the mean time downloading packages is speeded up anyway because the files are cached by the newly implemented Transparent Proxy.

I am still hopeful that the netboot/netinstall process can be implemented with the (modified) boot files being supplied by a local tftp server, but I'm no sure what, if anything, is being done to allow Macs boot and start the netinstall transparently, that is, not have to use OF to specify the tftp server explicitly from OF.

03Aug07

Shortcomings

• There is no PPC implementation of Automatix, the Ubuntu Customization application that allows users to easily add applications, third party and otherwise, that are not installed by default. The Automatix team dropped PPC platform support at the same time that Ubuntu moved PPC development from the release branch to the ports branch. FG Mac TechSupport will have to be prepared to offer alternatives to Automatix.
• Adobe does not provide a Flash player or plugin for the PPC platform, so web pages requiring a Flash plugin will not work. A noteable example is Youtube, but there are many others. There is an opensource alternative, Gnash, available in the repositories, but it is currently version 0.7.2. Version 0.8 is reported to support Youtube videos better, although Jeff Robinson told me that support is not complete. Version 0.8 is only available by downloading it from a GNU mirror.

07Aug07

G4 PCI
To quote Kyle and Stan (and others), I've learned something here today. Dan worked on a G4 last week that Ubuntu reported as a B&W G3, except for the fact that it had a G4 (7400) processor. The device assignments were like the G3 (hard drive is hdc, CD is hda), and the video card was in a PCI slot (no AGP slot at all). I mistakenly thought that this was some kind of home baked upgraded mutant. I was wrong. There were actually two versions of the G4 PCI released, a 350MHz and a 400MHz, and the logic board for these is essentially a B&W G3 logic board. These models were not manufactured for very long, and were shortly replaced with the AGP logic board models (named 'Sawtooth').

So what to do with these models in MacBuild? I told Dan to dismantle the one he was working on since I thought it was a mutant. I still think it's a mutant. I kind of get the impression that ultimately Apple thought it was a mutant, too. There aren't just a whole lot of them out there. It wouldn't be any great loss to simply recycle them, which is what I would recommend.

New User
A normal installation seems to populate the panels reasonably. We usually add a couple of things, a Terminal icon and a Gnome-PPP icon. If there is a DVD in the HW configuration, the local preferences are modified to use ogle instead of totem as the default video player. So if it were to come to pass that a new-user boot option were to be implemneted for Macs, how can these local mods in teh oem account be recreated in the new-user account? There should at least be a section in the PPC new user FAQs about how to make the mods manually. Perhaps the mods can be added with a custom package.

11Aug07

Tray Loader iMac
What if a volunteer wants to buy an iMac tray-loader, or a request for one appears on the wish list? What should MacBuild do about it, if anything?

Let's entertain the idea of providing such a thing for a moment. Several options present themselves. At the very least, the hard drive must be erased or replaced. That is possible to do if the CD works. Enough of these models still come in that a working CD replacement might be available. MacBuild may even have some now. The machines could be sold "as-is" for parts, or guaranteed to be working with OS 9 (not included). A working model in a more or less standard Apple configuration with OS 9 installed sold recently on ebay for about $67. The FG store price for that model would be $47. For parts ("as-is", no testing, wiped hd) I'm guessing the price would be about $16. For known good powerup, wiped hd, boots OS9), I'm guessing the store price would be about $24. These price guesses are strictly my own, and have not been sanctioned or even reviewed by FG store/sales staff.

Ubuntu, PowerPC, Entropy
In the course of refurbishing a QS/733 and a MDD/1.42GHz DP, and installing the Feisty (7.04) release, I discovered a couple of known bugs. On some later models of tower and laptop, gnome-ppp cannot discover the modem. It's weird. On the QS/733 the modem will be discovered, but on the MDD/1.42DP (and a G3/500MHz iBook) it will not. Anecdotal evidence from forums indicate that the modem works (wvdial succeeds), implying that the problem is with gnome-ppp and discovery, not with the modem or it's association with /dev/ttyS0. Unfortunately this means that the internal modem cannot be tested in this case.

There is another bug in which the CD player will freeze while playing a music CD. The default player used for testing is SoundJuicer. I tried playing a CD with gxine with the same result. There are also several posts about this on the web. I have not seen this problem with Feisty on g3 iMacs, but then again, I wasn't specifically looking for it. Maybe I'd better go back and check some iMacs.

Triage
The MacPile area, and the other connected shleves, are being dismantled to make way for a rearranged triage/eval/advanced-test space. There will be triage/eval benches where the shelves are now. I told Jeff that that I thought there would never be a need for mopre than two triage positions at such a bench. Jeff and Matteo suggested that a Mac Triage Flow description would be in order. There is already a first pass MacTriage description that I developed several months ago, based on the workspace arrangements/assumptions at the time. It's at User:Tonyr/MacTriage/23May07#iMacBuild. I'll use that as a starting point for a new description. There is also a build chart that details some of the steps at A_Mac_Rebuild_Table.

Here's the general idea. Triage still separates things we want to consider for rebuild (keeper) from things that we don't (reject). The basic criteria for 'keeper' are

• slot-load iMac, G3/G4/G5 tower, or any other desktop model in the G4/G5 generation that can be made to POST
• if iMac (all-in-one), displays video.
• any G3 Pismo or newer laptop

Before leaving triage for whatever reason, hard drives, memory and batteries are removed.

Anything else is rejected and marked for recycling, unless there is a special request for it.

The unresolved issue in this overview is what to do with rejected laptops. Rejected all-in-one iMacs can be stripped down to monitor-only status and recycled as such; there is informal precedent in Eval-1 for PC all-in-one machines (and Mac all-in-one Performas). Older Apple/Mac machines after mining are sent to recycling. PC laptops have their own workspace and process. Mac laptops are lumped together with all other Apple hardware. Currently, since all Apple/Mac hardware shares one process, old laptops are taken apart in the MacBuild area. Taking apart old laptops is not very difficult or time consuming, and can probably be done at the triage station.

The other unresolved issue is power supply reclamation and test. Power supplies should be reclaimed from otherwise rejected G3/G4 towers. It would be nice if power supply testing could be integrated into the existing power supply testing process, but I think it is too specialized. Apple has published testing procedures for the tower models, but they require that the power supplies be connected to an appropriate logic board and tested in situ, both with power off and power on. It would be necessary to set this up somewhere, and the existing power supply testing station may not be the best choice for a number of reasons.

15Aug07

Keyboard shorcuts
They are defined basically with gconf. There is a CLI tool for modifying gconf preference values, gconftools-2. The GUI tool is System->Preferences->Keyboard Shortcuts, but it is not very flexible. The Eject key is defined as '0xcc'. but I have no idea which key on the keyboard that is. The Eject key is defined in a schema defined in the file /usr/share/gconf/schemas/apps_gnome_settings_daemon_keybindings.schemas. The general problem is to get the F12 key to NOT be mouse-right-click, but be eject, and choose a set of keys for right-mouse-click and middle-mouse-click that are NOT F12 and F11.

Unmap F11/F12
This article http://ubuntuforums.org/showthread.php?t=468298 says uninstalling package mouseemu will free up the keys for redefinition.

Some say disable the Eject keyboard shortcut in System->Preferences->Keyboard Shortcuts.

Some say it's a conflict between pbbuttonsd and gnome keyboard shortcuts (disable shortcut suggestion). https://bugs.launchpad.net/ubuntu/+source/pbbuttonsd/+bug/91599

Some say look into the command powerprefs for control keyboard control management, but IIRC that article is from a few years back.

• Ref. 26Jul07 one button mouse entry.

Here's what I did, and it seems to provide reasonable behavior: 
1) removed package mouseemu with apt-get, 
2) edited /etc/sysctl.conf to comment out definitions for mouse keys, leaving only 3button mouse 
   emulation defined,
3) defined Eject to be F12 in System->Preferences->Keyboard Shortcuts

This allows the default definitions for right-click and middle-click to be Fn+Option and Fn+Apple.