|
andrew.mcmillan.net.nz
cd /var/www; more /dev/rant >>index.html
|
|
lca
A GPS is one of those toys that I have wanted for a very long time. So last year I finally marshalled enough excuses in one place, lined them up and plunked down some money for a Garmin GPSmap 60CSx in the vague belief that there's enough Linux software out there which understands GPS, and the Garmin is a brand that seems to have mature Linux support. And besides, I'd been told it was a good model that would do what I wanted and then some. Once I got it I naturally wanted to do stuff with it, and in particular I wanted to connect it up to my laptop which (of course) is running Linux and I found that it wasn't nearly as trivial as I had thought. This prompted me to run around finding software to use with it, so here's my capsule review of my journey from "extremely naive" to "very naive". Perhaps I'll also learn something from the comments of people who are further down the track. My first resort was "apt-cache search gps", of course, which immediately brings up such fine sounding programs as "gpsman", described as "A GPS manager" (I haven't managed to get it to do anything yet), "gpstrans" which since it specifically mentions Garmin GPS sounded like just the ticket (it appears to be quite old and superseded). There was some useful stuff as well.
Getting Maps on the GPSSince the Garmin GPSMap series will display maps, I wanted to be able to get some mapping data on there. There are two map sources I was interested in:
Using the NZ Open GPS MapsThe NZ Open GPS Maps are built specifically for the Garmin GPS, so getting them onto my GPS was relatively simple once I found out the exact command line was remarkably hard. The program that I needed to find to be able to get the maps onto my GPS was I downloaded the maps from the NZ Open GPS Maps hosted on the cGPSMapper site. I needed the files identified as 'binary' files - the installer is a Windows program and no use on Linux. They all have filenames like The command-line I first used to install the maps onto the GPS was: That works fine, but if you have even a moderately sized micro SD card in the GPS there are much faster ways. As I found more free maps I found much better ways to do it. An alternative approachIn the GPS (well, in my one anyway) there is a micro-SD card which is FAT formatted and all of the installed maps are in one humungous combined image of all of the uploaded maps. This means that there is no way to straightforwardly add/remove maps without creating that image, and re-uploading the whole thing. The image file is called 'gmapsupp.img' and is in the 'garmin' subdirectory on the SD card. This means that you can create the file (or even several different files, with different maps on them) and move them directly on there much quicker, either by taking the SD card out and using it in a USB2 reader (for really big files) or switching the GPS to operate in USB Storage mode (which is USB1, but OK for smaller files). You can create the IMG file to copy in as QLandKarteAnother program I found useful specifically for dealing with the Garmin GPS Map format is QLandKarte, which understands the native Garmin format and will display the maps in a graphical window. You can also point and click to select maps, building up a specific set that you can then upload to the device from within QLandKarte itself. This seems to operate a lot faster than sendmap20, presumably because it's driving the USB directly, rather than through usb_serial. To get it to work I seem to have to (a) sudo rmmod garmin_gps and (b) sudo QLandKarte, however, which is definitely not ideal. For the moment I will continue to use sendmap20 to create files which are sets of the maps I want, and will put them on the device in USB storage mode. QLandKarte is still very useful, however, for looking at maps and deciding if they are worth bothering with. I used it in this way to select the Australian maps I took with me when I went to Linux.Conf.AU recently. There are more free Garmin GPS maps. OpenStreetMap.orgThose Australian maps came from OpenStreetMap.org and ultimately I expect that it will become the best source of data for creating maps for my GPS. OpenStreetMap.org is an attempt to provide community-maintained maps for the whole world and which seems to have made significant progress in UK, Europe, America and Australia as well as many other parts of the world. The data from OpenStreetMap.org does not have the same licensing restrictions as are present even in the NZ Open GPS Maps data (which is relatively free, but has problems similar to the old-style BSD license). There are also some good mapping interfaces to the OpenStreetMap data although the searching currently still leaves something to be desired. Interfaces are also available for extracting subsets of the data, which is in a fairly straightforward XML format, or you can download the whole lot, but at nearly 3GB for the bzip2 compressed version you won't do it every day. I gave a brief overview talk about OpenStreetMap.org while I was at Linux.Conf.AU and it seemed to go down quite well (I gave it several times, in fact). You don't need a GPS to contribute to the project, and of course the maps which are being created are usable for many purposes beyond their usefulness on GPS. Some people apparently even print them out, but that just seems weird! LCA officially started today, with a great opening keynote by Bruce Schneier very nicely pointing out the differences between feeling secure and being secure, and with a few good examples of where these are mismatched and action needs to be taken to fix the mismatch. Good stuff - catch it on video if you get the chance. Of course for me LCA started on Monday when I got roped into presenting a lightning talk at the Community Wireless miniconf on OpenStreetMap.org which is something I've been fiddling with recently, and a project which definitely deserves more attention. Since nobody else turned up to give their lightning talks my 3 minute presentation was stretched severely out of shape, but the Community Wireless miniconf was generally pretty interesting and I stayed there all day in the end. Yesterday was the SysAdmin miniconf, and as one of the organisers there was no chance for me to look around at what else was going on. It was a full-on day, and the feedback I have from everyone is that it was a good miniconf. Certainly it was good enough to attract more people than could fit in the 120 seat theatre we initially had allocated so that we had to move into a much larger theatre. A big thank-you to the organisers for so quickly recognising and reshuffling things so that was possible. All signs so far definitely indicate another great linux.conf.au Opening the box, we find...Inside the box was a computer, a power adapter and a battery. There was a bunch of other stuff, but none of that applied to me! I booted from an Etch pre-release CD I had floating around, most things were detected and most things worked in a basic manner. I don't run Etch though, so I was upgrading to the truly bleeding edge fairly rapidly. Normally I run Debian Unstable, so I upgraded to that immediately and that was all fine, but with Etch being released soon many developers are putting interesting stuff into Experimental, and I wanted X.org 7.2 and Gnome 2.16 which are in there... Packages you will want to installTo get the 1400x1050 screen resolution you will need to install '915resolution'. This seems to still be the case with the packages of X.org 7.2. I have seen some interesting sounding options in the i810 man page that might be alternative ways, but the hack that 915resolution uses is a much simpler solution since it is already packaged. To get the wireless working you will need to install 'ipw3945d' and 'ipw3945-modules-2.6.18-4-686' (or some other package of the modules - I installed the source package and used module-assistant to build the modules because I like to build my own kernel). What is mostly workingA Tour on the PCI Bus
DisplayOnce I installed 915resolution the display just works under X.org 7.1 and X.org 7.2. After a suspend to disk I have to run 915resolution again otherwise X will be shown in 1024x768. I think I could then run xrandr to switch back to 1400x1050 at that point (to be confirmed). TouchpadThe touchpad has a scrolling area at the right of it, which works as designed, and it also seems to handle a two-finger tap as equivalent to a two-button click (i.e. what would normally be a middle button click). It seems nice, and I will probably get used to it. PCMCIAI pulled out my 3G card and plugged it in and it just works. There is also one of these new fandangled PCMCIA^WCardBus replacement whatsits, but I don't anticipate having anything that actually uses that any time soon. USBAll USB ports seem to work as expected. I seem to get pretty high speed transfers out of my camera too. SD * MS/Pro * MMC * SM * XDThis was something that I was hopeful would actually work, since I have a few devices that use SD cards now and it definitely seems to be the format that is taking over from Compact Flash, except for in irrelevant stuff like Sony and Olympus equipment. Unfortunately, though I compiled all of the MMC kernel stuff into my kernel (as well as the core under "Device Drivers -> MMC in the kernel configuration there are chipset drivers under "Device Drivers -> Misc devices" - go figure ). Those drivers seem to be for a "TI PCI74xx compatible" reader, and mine seems to be a "PCIxx12" something, so perhaps there just isn't a driver Right Now and I just have to wait a few months. Update: 2007-02-23 What Ain't TestedBluetoothI can see the bluetooth is there with 'lsusb', and I expect that it will all work just fine, but I don't have anything that does bluetooth, so I can't say for sure. ModemI don't have a poorband account anywhere, so I can't test it. It isn't something I expect to ever want to use, but it looks like an AC97 modem so I expect it will work just fine. TV OutSince I don't have a television. Parallel PortDidn't these things get replaced by network printers and USB printers a bunch of years ago? Serial PortYep. It has one. Weird. I wish that they had put two USB ports there instead - it would have been a much better location for stuff like my wireless mouse to plug into. FirewireI must pull one of my firewire HDs out and see if this works... What Ain't WorkingSuspend / ResumeSuspend to ram sometimes works for me, but on most occasions resume has left me with a working system but no keyboard input. Some googling suggests there are ways around this, but I am not sure if it is fully reliable at this stage. I have tried with the 'irqpoll' option included on the kernel boot line, which is purported to help with other laptops, but this (a) doesn't help, and (b) lots of other stuff seems to work badly with it. Somewhere I read a suggestion to use the 'noapic' boot option, but this won't boot. I have a vague memory that the APIC is actually required for SMP systems. Suspend to disk works, except that X comes back without the BIOS manipulations that have been done by 915resolution, so it manages to end up in a 1024x768 screen rather than 1400x1050. This can probably be fixed by also running 'xrandr' to set the resolution back after '915resolution' has been run, or by 'chvt 1' prior to suspend, then '915resolution' on resume before 'chvt 7' to bring the X screen back. I will see... Update: 2007-02-23 Suspend to disk works reliably for me now that I "chvt 1" ... suspend to disk ... "/etc/init.d/915resolution start; chvt 7". If the keyboard is not working after a resume from ram I can plug in a USB keyboard and that does work, but unloading and reloading the atkbd, psmouse and libps2 does not help (as I had hoped it would). This is really important to me, so I'll keep working on it until I can either workaround the problems, or it gets to be supported by 2.6.21 or so... Hopefully sooner rather than later! Fingerprint ReaderWell, I honestly wasn't expecting to have much luck with the fingerprint reader, and I haven't. A bit of googling did throw up Michael Crusoe's Bio API implementation which I have been unable to get to build so far, as well as the proprietary driver for the UPEK fingerprint reader, which refuses to install, probably because the Bio API isn't installed... Looking at the output of 'lsusb' it seems that it may well not be a UPEK device, since I am told "Bus 001 Device 004: ID 08ff:2580 AuthenTec, Inc.". More investigation is clearly required. That is all more than I expected to be there, to be honest, but unfortunately not enough to get me going. Oh well, I will have to take some heart from the fact that these devices are cute, rather than fully trustable. If I do eventually get it working, I now know that I need to revisit this setting up the fingerprint reader page, which seems comprehensive. Overall Impressions
|
|