Greg
Greg's diary
June 2005
Translate this page
Select day in June 2005:
Su Mo Tu We Th Fr Sa
1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30
Select month:
2005 Jan Feb Mar Apr
2005 May Jun Jul Aug
2005 Sep Oct Nov Dec
Today's diary entry
Diary index
About this diary
Previous month
Next month
Greg's home page
Greg's photos
Network link stats
Greg's other links
Copyright information
    
Groogle

Wednesday, 1 June 2005 Echunga
Top of page
next day
last day

Somehow I didn't get anything done today: lots of little things got in the way, though it's difficult to understand how they managed to take up the entire day.

Round midday an evaluator from the insurance came to look at the remains of our air conditioner. Summary: nothing wrong with the compressor. He put 1000V across it and got no leakage, so it's unlikely to be in too bad shape. Looks as if Phil measured some leakage current (it was 20 MΩ to earth, which I noted at the time was not going to cause much overheating). So no new air conditioner right now.

Apart from that, spent an inordinate amount of time doing an expense report. Why does this take such a long time?


Thursday, 2 June 2005 Echunga
Top of page
previous day
next day
last day

Still little rain. The Bureau of Meteorology states in the May rainfall bulletin:

  Mt.  Lofty Ranges (23C)
                             Rain (mm)        Rain Days   Decile
                           Obs  |  Ave       Obs | Ave

              ECHUNGA               94.3            12
              MEADOWS       3.4     99.3       1    13        L
      

The values for Echunga are typical: the Bureau of Meteorology seems to miss out most measurements, and so we don't know what fell there. Even with Meadows, I'm not convinced that they accounted for all rainfall. Still, 3.5% of the monthly fall is far below normal, and we're still hurting. I think the L at the end of the line stands for “lowest on record”.

Spent some time trying to set up a pf firewall. It's not made easier by the documentation, which is copious but unstructured. pf seems to offer lots of unrelated additional ideas, such as SYN proxy and anti-spoofing. They look like a good idea, but it's difficult to know when you're done.


Friday, 3 June 2005 Echunga
Top of page
previous day
next day
last day

Somehow I'm still not getting much work done. Spent more time looking at pf, and the more I look, the more I think changing over to it should be independent of the ADSL installation. I'm still confused about this GRE/IP installation, all the more so since Daniel O'Connor confirmed that there's a third address involved:

 gif0: flags=8051<UP,POINTOPOINT,RUNNING,MULTICAST> mtu 1452
         tunnel inet 150.192.32.134 --> 203.25.129.227
         inet6 fe80::240:c7ff:fe99:3a7c%gif0 prefixlen 64 scopeid 0x6
 dc0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
         inet 203.91.11.1 netmask 0xffffffc0 broadcast 203.91.11.63
         inet6 fe80::240:c7ff:fe99:3a7c%dc0 prefixlen 64 scopeid 0x1
         inet 203.91.11.4 netmask 0xffffffff broadcast 203.91.11.4
         ether 00:40:c7:99:3a:7c
         media: Ethernet autoselect (100baseTX <full-duplex>)
         status: active
 dc1: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
         inet 150.192.32.134 netmask 0xfffffffc broadcast 150.192.32.135
         inet6 fe80::240:c7ff:fe9a:1420%dc1 prefixlen 64 scopeid 0x2
         inet 192.168.1.2 netmask 0xffffff00 broadcast 192.168.1.255
         ether 00:40:c7:9a:14:20
         media: Ethernet autoselect (10baseT/UTP)
         status: active
      

To protect Daniel, these aren't the real addresses, though they're similar. It's clear that the external address of the tunnel is 150.192.32.134, and the dc0 interface address is the tunnelled /26 network, but it's not clear why 203.25.129.227 is needed. Is it one of these dummy addresses like the 10.0.0.1 that I used on my satellite downlink?

The answer might have come in the afternoon, when I got a call from Ben at Internode, but unfortunately I was in town at a Rocksoft technical meeting—the first in 6 weeks—so had to put it off until after that.

After that meeting was another discussion on file systems, which dragged on until 5 pm, making it impossible to call Ben back. The discussion was on the same things we were discussing 6 weeks ago. We don't seem to have made much headway.

Back home and found a message from Peter Hansteen, who read my comments on the lack of overview documentation for pf, and reminding me that he had sent me some for review some months ago. Egg on my face. It's amazing how easily you forget these things if you don't need them at the time.


Saturday, 4 June 2005 Echunga
Top of page
previous day
next day
last day

Slept for 12 hours last night. I really seem to be tiring myself too much; I have barely stopped rotating for 2 months. Today decided to take it easy and just play around with multimedia stuff, in the process discovering just how really impossibly bad the general quality of the software is. For example, xawtv gets completely confused by dual-head displays such as the monitor/TV combination on teevee.lemis.com. The TV display is :0.1, and if I try to run xawtv on it, it displays a broken image on the root window of :0.0. If I try to start it on :0.0, it doesn't give me the menu items for tuning. I can get it to display correctly on tvremote.lemis.com:0.0, but of course the network isn't fast enough to keep up a normal frame rate.

On the positive side, did manage to get my tuner card working, by trial and error. It appears to be type 13. I'm really not happy that I couldn't find it out any other way, though.


Sunday, 5 June 2005 Echunga
Top of page
previous day
next day
last day

Another quiet day. Went riding again—this is supposed to be a regular Sunday thing, but since its inception some time in March it's only the second time—and apart from that did little. Tried to find a way to get mplayer to play from a tuner, but again the ambiguity in the documentation made it almost impossible. It seems that you need to build mplayer after installing tuner software; what a pain! Did some more investigation and found another program suite, bsdav, which might work better, though I have my doubts about compatibility. Took the easy way out and did nothing.


Monday, 6 June 2005 Echunga
Top of page
previous day
next day
last day

Called up Ben at Internode this morning to discover that my ADSL line had already been provisioned—but that my router had been entrusted to Australian Air Express, who had already managed to make a mess of a delivery from Dell in January (they delivered an overnight parcel to the Echunga post office, and refused to pick up a prepaid pickup parcel). It seems that they're up to their usual tricks, and I was told it had been offloaded to Australia Post. Anyway, it wasn't there today, though Ben is confident (probably as the result of lack of experience) that it'll be there tomorrow. Damn. I could easily have gone into town and picked on up, but obviously they don't want to give me a second one.

Apart from that, work on my debugging tutorial. I need to re-wrap it as a userland tutorial, and spent some time looking at the Makefile from hell.


Tuesday, 7 June 2005 Echunga
Top of page
previous day
next day
last day

Started off working on my documentation this morning, until we were able to check whether the ADSL router (sent from Adelaide on Friday via Australian Air Express) had arrived at Echunga post office. Of course, it hadn't. What use are these couriers if they first require me to go to the post office (thus merely shortening my car journey from 40 km to 5 km) and then can't do it in two business days? Called up Internode and arranged to pick up another unit.

Back home with the new router and plugged it in: instant gratification. But I still had this GRE tunnel to contend with, and try as I might, I couldn't get it to work. That's at least partially because of the absolutely appalling documentation about it: nobody can give me clear instructions on how to configure the tunnel and how to set up the routing. Frustrating day.


Wednesday, 8 June 2005 Echunga Images for 8 June 2005
Top of page
previous day
next day
last day

Up early this morning to investigate why my tunnel wouldn't come up. Spent several hours, comparing closely Daniel O'Connor's configuration. I wish I understood this better.

After several hours, got on to Ben at Internode and discovered we had the wrong type of tunnel. They had set up a GRE tunnel, apparently what they normally use for Linux, but the configurations that both Ben and Daniel had given me (using the gif interface) uses IP-IP tunnels. Ben changed the tunnel type and it worked perfectly. Just about the first thing I did was to make a change to the man page gif(4):

--- gif.4       6 Dec 2004 19:31:35 -0000       1.24
+++ gif.4       8 Jun 2005 01:53:25 -0000       1.25
@@ -27,7 +27,7 @@
 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 .\" SUCH DAMAGE.
 .\"
-.\" $FreeBSD: src/share/man/man4/gif.4,v 1.24 2004/12/06 19:31:35 glebius Exp $
+.\" $FreeBSD: src/share/man/man4/gif.4,v 1.25 2005/06/08 01:53:25 grog Exp $
 .\"
 .Dd April 10, 1999
 .Dt GIF 4
@@ -50,6 +50,11 @@
 .Nx ,
 .Nm
 can also tunnel ISO traffic over IPv[46] using EON encapsulation.
+Note that
+.Nm
+does not perform GRE encapsulation; use
+.Xr gre 4
+for GRE encapsulation.
 .Pp
 Each
 .Nm
@@ -183,6 +188,7 @@
 .Va net.link.gif.parallel_tunnels
 to 1.
 .Sh SEE ALSO
+.Xr gre 4 ,
 .Xr inet 4 ,
 .Xr inet6 4 ,
 .Xr ifconfig 8
      

So what's the difference between GRE and IP tunnelling? It seems that GRE adds a four-byte header in front of the payload:

 
This should be GRE-detail.gif.  Is it missing?
Image title: GRE detail
Dimensions: 434 x 170, 4 kB
Dimensions of original: 434 x 170, 4 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Wednesday, 8 June 2005:
thumbnails    small images    diary entry

This header is missing in the IP encapsulation. Here are dumps of a typical GRE and IP packet:


This should be GRE.gif.  Is it missing?
Image title: GRE          Dimensions:          902 x 1134, 256 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Wednesday, 8 June 2005, thumbnails          All images taken on Wednesday, 8 June 2005, small
Diary entry for Wednesday, 8 June 2005

 
This should be IP.gif.  Is it missing?
Image title: IP          Dimensions:          1402 x 1174, 224 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Wednesday, 8 June 2005, thumbnails          All images taken on Wednesday, 8 June 2005, small
Diary entry for Wednesday, 8 June 2005

 

This obviously makes IP encapsulation a (marginally) better choice for this kind of application.

With that, things worked relatively well, and I was able to connect the ADSL link via the same interface. This is one advantage of the tunnel approach: since the only traffic down the tunnel is encapsulated, it all goes through the machine with the gif interface, and it can be firewalled there. I'm still refining things, but the situation currently looks like this:

With all that, the configuration looks like:

xl0: flags=8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> mtu 1500
        options=9<RXCSUM,VLAN_MTU>
        inet6 fe80::250:daff:fecf:735%xl0 prefixlen 64 scopeid 0x1
        inet 192.109.197.137 netmask 0xffffff00 broadcast 192.109.197.255
        inet 150.101.14.10 netmask 0xfffffffc broadcast 150.101.14.11
        ether 00:50:da:cf:07:35
        media: Ethernet autoselect (100baseTX <full-duplex>)
        status: active
lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> mtu 16384
        inet 127.0.0.1 netmask 0xff000000
        inet6 ::1 prefixlen 128
        inet6 fe80::1%lo0 prefixlen 64 scopeid 0x3
tun0: flags=8151<UP,POINTOPOINT,RUNNING,PROMISC,MULTICAST> mtu 1500
        inet 139.130.136.138 --> 139.130.136.129 netmask 0xffffffff
        Opened by PID 38825
gif0: flags=8051<UP,POINTOPOINT,RUNNING,MULTICAST> mtu 1452
        tunnel inet 150.101.14.10 --> 203.16.215.227
        inet6 fe80::250:daff:fecf:735%gif0 prefixlen 64 scopeid 0x7

Routing tables

Internet:
Destination        Gateway            Flags    Refs      Use  Netif Expire
default            150.101.14.9       UGS         0   309339    xl0
127.0.0.1          127.0.0.1          UH          0   102343    lo0
139.130.136.129    139.130.136.138    UH          0        5   tun0
150.101.14.8/30    link#1             UC          0        0    xl0
150.101.14.9       00:90:1a:40:09:98  UHLW        1        3    xl0    531
150.101.14.10      00:50:da:cf:07:35  UHLW        0        6    lo0
150.101.14.11      ff:ff:ff:ff:ff:ff  UHLWb       0      287    xl0
192.109.197        link#1             UC          0        0    xl0
      

I've left out some irrelevant entries in the list above.

I haven't got round to rebooting the machine (I hate doing that!), and in this case I didn't even need to take down many network connections (only the ones that went directly over the PPP link), so the following entries in /etc/rc.conf haven't been tested:

# ADSL tunnel
ifconfig_xl0_alias0="inet 150.101.14.10 netmask 0xfffffffc"
gif_interfaces="gif0"
gifconfig_gif0="150.101.14.10 203.16.215.227"
ifconfig_gif0="mtu 1452 up"
defaultrouter="150.101.14.9"
      

After that, addressed the router configuration: somehow I ended up with an invalid password, though nobody had changed it, so followed the minimal instructions in the sheet of paper that came with the router and reset it—and the configuration. As a result, I was off the net until I could find somebody to help me set it up, not made any easier by the fact that I had been told that the router should operate in RFC 1483 bridging mode, but in fact it was in “bridge only” mode. One day I'll find out what the difference is, but I suspect that “bridge only” means “like an expensive variant of an ADSL modem”. Here's the configuration:


This should be config.gif.  Is it missing?
Image title: config          Dimensions:          822 x 1020, 102 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Wednesday, 8 June 2005, thumbnails          All images taken on Wednesday, 8 June 2005, small
Diary entry for Wednesday, 8 June 2005

 

After that, the next step was to change my routing away from Telstra. Given the trouble I had last time, I expected that to be difficult, but in fact they sorted things out within an hour. After that, I had everything on ADSL, right? Wrong. It seems that Epoch are still advertising the satellite route, so a lot of my traffic is, once again, coming in via satellite. Since the PPP line still isn't dead—I still have an MX record pointing to that interface, originally in case the satellite link went down—I'm still getting very persistent spam attempts from somewhere in Russia, it seems:

07:30:52.875598 IP 82.204.224.110.23174 > gregl1.lnk.telstra.net.smtp: S 112858842:11285
8842(0) win 57344 <mss 1448>
07:30:54.265315 IP 82.204.224.110.23156 > gregl1.lnk.telstra.net.smtp: S 3661548368:3661
548368(0) win 57344 <mss 1448>
      

And this despite the fact that I've firewalled them off.

Apart from that success, another: it's raining! At least two months late, it has finally started to rain. I'm reminded of the jubilation evident in C. J. Dennis' “A Song of Rain”.

In the afternoon, a phone call from Bernd Wulf, who wanted to tell me that he had just changed from satellite to ADSL. His experiences were somewhat different, though: while investigating getting a connection with a different provider, Telstra told him that all of his phone lines were on pair gain, so he couldn't get ADSL. Then, just after the DSLAM was installed, and before other providers were informed, Telstra contacted him and said that yes, indeed, one of his lines was available. So he signed up with Telstra and filed a complaint with the Australian Consumer and Competition Commission. That's very different from my experience: I've tried to sign up with Telstra earlier, and basically been told “Don't call us, we'll call you”. Well, they haven't yet. I wonder how long it will take them, if they notice at all.


Thursday, 9 June 2005 Echunga
Top of page
previous day
next day
last day

So now the network is up and running, though I have no less than three downlinks. In the afternoon, changed that. It's something like the passing of an era: I started using modems in about 1980, at the time with a steam-driven 300 bps acoustic coupler. This afternoon I stopped my last PPP link and thus stopped using modems. Fittingly, the last messages came from a particularly persistent spammer:

14:37:42.441281 IP 82.204.224.110.22445 > gregl1.lnk.telstra.net.smtp: S 2164515201:216
4515201(0) win 57344 <mss 1448>
14:37:43.061400 IP 82.204.224.110.22491 > gregl1.lnk.telstra.net.smtp: S 3912128432:391
2128432(0) win 57344 <mss 1448>
14:37:49.260078 IP 82.204.224.110.22491 > gregl1.lnk.telstra.net.smtp: S 3912128432:391
2128432(0) win 57344 <mss 1448>
14:37:53.129350 IP 82.204.224.110.22486 > gregl1.lnk.telstra.net.smtp: S 301016964:3010
16964(0) win 57344 <mss 1448>
14:38:00.508024 IP 82.204.224.110.22471 > gregl1.lnk.telstra.net.smtp: S 2443297115:244
3297115(0) win 57344 <mss 1448>
      
My firewall shows that he had tried over 13,000 attempts in the previous 12 hours:
06000    13564     720168 unreach port log logamount 100 tcp from 82.204.224.110 to any
      

Getting the satellite disconnected was less easy. Called up ihug, who told me to talk to BorderNET, who told me to talk to ihug. Finally got BorderNET to realize that they were probably breaking the law by not letting go of my data, and they gave me a number in New Zealand, +64 9 3592730, and the name Willy Prinsloo. The only problem is that nobody answered. Called ihug again, and they told me that this was an administrative contact, not a technical one, and that nobody knew a technical contact. What a mess! Daniel of ihug did try to stop the service, but it didn't seem successful.

Maybe that was just as well: it's not clear how well Internode's advertisement of my net block is progressing. On IRC, discovered route-views.routeviews.org, who provide a telnet-based BGP service. Trying that showed:

Username: rviews
route-views.oregon-ix.net>sh ip bgp 192.109.197.0
BGP routing table entry for 192.109.197.0/24, version 61511243
Paths: (49 available, best #32, table Default-IP-Routing-Table)
  Not advertised to any peer
  286 209 3561 4565 7657 7657
    134.222.85.45 from 134.222.85.45 (134.222.85.45)
      

The first numbers under “not advertised” are the AS numbers to be traversed to access the net block. AS7657 is The Internet Group Ltd, NZ, who appear to be behind ihug. Internode is AS4739, and it doesn't show. Tried to contact Internode, but it seems that Ben was off sick today, and Mark Newton didn't come on line until later. He says it's fixed now.

Apart from that, down to Stirling to meet up with Jim Johnson at the same cafe where we met in March last year. Interesting developments are on their way.


Friday, 10 June 2005 Echunga
Top of page
previous day
next day
last day

Into town this morning to present a (userland) debugging seminar to people at work at 10 am. Got there and found nobody else there; everybody was late, and two of the key attendees were off sick. In addition, I had asked for a reproducible and real bug in our software, but somebody had gone and fixed them all. Still, the presentation (more like a demonstration) went well, and there was a good level of audience participation.

Meeting in the afternoon, and we're still going over the same issues we were nearly 2 months ago. I suppose we've made some progress, but it should be much more.

Brought home the data projector, a Sony VPL-ES2, from work in the evening. I've been thinking about buying one, but first I wanted to find out what it's really like. A good thing, too: I found a number of things, some of them unexpected:


Saturday, 11 June 2005 Echunga Images for 11 June 2005
Top of page
previous day
next day
last day

Slow day. In the morning, planned to bottle beer, but after one brew, discovered I would have nowhere to put the second brew: looks like I'm drinking less beer than a year ago. I wonder why.

Spent some time looking at the projector, and got further. Found a VGA extension cable—unfortunately only 2 metres—and connected up the projector directly to teevee.lemis.com, which I had to move into view (and earshot) for that purpose:


This should be Lounge-1.jpeg.  Is it missing?

Image title: Lounge 1          Dimensions:          2048 x 1536, 352 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Saturday, 11 June 2005, thumbnails          All images taken on Saturday, 11 June 2005, small
Diary entry for Saturday, 11 June 2005 Complete exposure details

 

The result is more practical than pretty, and much needs to be done, but at least it works. Both teevee and the projector made far too much noise, and the position of the projector and cable in the way are both unacceptable in the long run. A longer VGA cable would solve the problem with teevee, but I don't know what to do about the projector. A quick search on the web confirmed that there are some quite reasonable 1024x768 projectors available out there—the BenQ PB6210 looks like a good candidate—but they all advertise things like “60"@2m wide-angle lens”. Apart from the amusing confusion of units (why not “1.5m@75"”?), this is exactly the problem that I'm facing with the Sony projector. So what was intended to be an advantage (it's much more difficult to build a lens with such a wide angle) has become a disadvantage. More research needed.


Sunday, 12 June 2005 Echunga
Top of page
previous day
next day
last day

Brew day today. Since it's winter, and I have enough beer already, decided to try an approximation to a Pilsener. That started with disaster, reminding me of a couple of German proverbs: “Daran sind Hopfen und Malz verloren” (“Hops and malt are lost on that”) and “Das ist für die Katz'” (“That's for the cat”, i.e. no longer useful for human consumption). What happened was that, unusually, I poured the malt straight into the mash water without weighing it—and without seeing that it hadn't been crushed. It was too late to crush it then, of course, (“Daran ist Malz verloren”), and it wouldn't mash like that, so all I could do was leave it to soak and give it to the horses (“Das ist für das Pferd”).

Down to Grumpys and got some more malt. Though there's no evidence that it was their fault, I got the replacement malt for free. Nice to have a place like that nearby. Back home and had a fairly normal brew day. I think one brew is so much less stress than two, so I should do single brews more frequently.

Watching TV on the projector in the evening. It's going to be difficult to go back to a tube. Noted another irritating thing about the Sony projector: it has a brightly lit “SONY” emblem on the back. That serves no good purpose, and it's irritating. What good does that kind of advertisement do?


Monday, 13 June 2005 Echunga Images for 13 June 2005
Top of page
previous day
next day
last day

Today was one of the Queen of England's many birthdays. If she really became a year older every time her birthday is observed somewhere, she'd be over 1000 years old by now.

Tidied up the lounge room in the morning, partially as a result of seeing the photo I took yesterday, and discovered many magazines that I had received but not yet read. The oldest was dated December 2003, so spent the rest of the day reading magazines and trying to decide about video projectors. Also took some photos of Yvonne teaching Darah to take the bit by herself. We still have a way to go.


Tuesday, 14 June 2005 Echunga
Top of page
previous day
next day
last day

Getting back to work after a long weekend isn't easy, especially when a lot of the work is administrative. Finally cancelled two phone lines (Yana's old line and the dial-up line I used to use for the Internet connection) and also the dialup connection itself, ending nearly 8 years of connectivity via Telstra. Also got the bill for the last month for that connection: $293.45 for rather over 1 GB of data. I had forgotten how expensive Telstra's “Big Pond Direct” service is: it must be the only Internet connection in the world which is more expensive now than it was 8 years ago. Still, that's the last of those bills.

I'm still getting traffic routed via satellite. Contacted the iHug NOC in New Zealand, and they wanted email from me to cancel the routing. We'll see if that works.


Wednesday, 15 June 2005 Echunga
Top of page
previous day
next day
last day

Only 8 days ago we were in the middle of one of the worst droughts on record. Now it's like all the rain that we missed at the time has arrived at once. We must have had 100 mm in the last week, and it's showing no sign of letting up. To add to the situation, one of our gutter drain pipes has blocked up, and there was water all over the place. Looks like a bit of maintenance will be needed tomorrow.

More design work today. I've done much of this work a couple of months ago, but somehow finding a good way to express it kept me busy all day.

I'm still getting traffic routed via satellite. Contacted Erick at the iHug NOC people in New Zealand again (phone +64-9-359-2708). He told me that they didn't receive the mail until after they finished working for the day, but that he would do something today. Strange hours they keep at the NOC, though:

Jun 14 11:59:16 wantadilla postfix/smtp[78341]: 3E1E485678: to=<noc@ihug.co.nz>, rel
ay=zuul.ihug.co.nz[203.109.252.44], delay=12, status=sent (250 ok:  Message 351518968 acce
pted)
      

There are 2½ hours time difference between here and New Zealand, but that means that the message must have got there at 2:30 pm.


Thursday, 16 June 2005 Echunga
Top of page
previous day
next day
last day

I'm still getting traffic routed via satellite. Today tried a different tack and contacted Daniel in Sydney. Daniel's in admin, but he tried to get me disconnected last week. Possibly he'll get things sorted out better.

Into town to my tax advisor round lunch time, rather earlier than I intended, but it turned out to be a good thing: after the meeting I had to go out to Wingfield to collect the fan for the air conditioner that had been taken for repair at the beginning of the month, and then to the other side of town to buy some cables. For some reason, everybody was driving at a snail's pace, and I made it to the ICT Council meetings only just in time.

The meetings were lively, and went on for over 4½ hours. Despite the length, they seemed to have been useful, but it made for a tiring day.


Friday, 17 June 2005 Echunga
Top of page
previous day
next day
last day

Came into the office this morning to discover that I was still getting traffic routed via satellite. Tried the iHug NOC people in New Zealand again and got hold of Jules, who did something about it immediately. It turned out that they had done something already. The routing tables looked different:

route-views.oregon-ix.net>sh ip bgp 192.109.197.0
BGP routing table entry for 192.109.197.0/24, version 65919052
Paths: (51 available, best #4, table Default-IP-Routing-Table)
  Not advertised to any peer
  2905 701 2914 4739
    196.7.106.245 from 196.7.106.245 (196.7.106.245)
      Origin IGP, metric 0, localpref 100, valid, external
  6079 2914 4739
    207.172.6.162 from 207.172.6.162 (207.172.6.162)
      Origin IGP, metric 6, localpref 100, valid, external
  267 2914 4739
    204.42.253.253 from 204.42.253.253 (204.42.253.253)
      Origin IGP, metric 0, localpref 100, valid, external
      Community: 267:2914 2914:410 2914:2402 2914:3400
  2914 4739
    129.250.0.11 from 129.250.0.11 (129.250.0.88)
      Origin IGP, metric 256, localpref 100, valid, external, best
      Community: 2914:410 2914:2402 2914:3400
      
4739 is Internode, so that looks good. Further down, though, we found:
  11608 4565 7657 7657
    207.246.129.14 from 207.246.129.14 (207.246.129.14)
      Origin IGP, localpref 100, valid, external
      Community: 11608:444 11608:801 11608:1008 11608:6601
      

And 7657 is iHug. Jules said that he would check the server in the USA, but that he had lost the password, and it might take a while. It appears that he was successful, though:

08:35:19.522359 < 204.152.190.13.ssh > 192.109.197.135.61403: P 2908400:2908448(48)
ack 1729 win 33580 <nop,nop,timestamp 4103124 647965820> [tos 0x10]
08:35:19.529443 < 204.152.190.13.ssh > 192.109.197.135.61403: P 2908448:2908608(160)
 ack 1729 win 33580 <nop,nop,timestamp 4103124 647965820> [tos 0x10]
08:35:19.529540 < 204.152.190.13.ssh > 192.109.197.135.61403: P 2908608:2908656(48)
ack 1729 win 33580 <nop,nop,timestamp 4103124 647965820> [tos 0x10]
08:35:19.529623 < 204.152.190.13.ssh > 192.109.197.135.61403: P 2908656:2908704(48)
ack 1729 win 33580 <nop,nop,timestamp 4103124 647965820> [tos 0x10]
08:35:19.529725 < 204.152.190.13.ssh > 192.109.197.135.61403: P 2908704:2908752(48)
ack 1729 win 33580 <nop,nop,timestamp 4103124 647965820> [tos 0x10]
08:36:00.806559 > 10.0.0.1.who > 10.255.255.255.who: udp 132 (DF)
08:39:00.806551 > 10.0.0.1.who > 10.255.255.255.who: udp 132 (DF)
08:42:00.806561 > 10.0.0.1.who > 10.255.255.255.who: udp 132 (DF)
08:45:00.806553 > 10.0.0.1.who > 10.255.255.255.who: udp 132 (DF)
08:48:00.806559 > 10.0.0.1.who > 10.255.255.255.who: udp 132 (DF)
      

So after nearly 4 years, my satellite service is over as well, and I'm finally only on ADSL. That took some doing.

Chris Yeardley along in the evening, bringing La Tigre with her.


Saturday, 18 June 2005 Echunga
Top of page
previous day
next day
last day

Quiet day. The rain continues, and there's not much we could do outside. Spent some time looking at a sorely neglected AUUGN instead. Most of the content is my own; the rest is by Frank Crawford and John Lions.

In the evening talking to Chris Yeardley about networking. She bought a satellite dish from me about 18 months ago, but never got it working. It's a bit late now, anyway. Talked about ADSL, and had a surprising amount of difficulty finding Telstra's ADSL Demand Register; here it is now for future reference. The trouble is that you still need to know the name of the exchange; the corresponding phone number query doesn't help very much if the phone hasn't been installed yet.


Sunday, 19 June 2005 Echunga Images for 19 June 2005
Top of page
previous day
next day
last day

Yvonne had the chance to get some photos of La Tigre in an American magazine, but the deadline is tomorrow, so despite the inclement weather spent most of the morning trying to get some good photos of here, ultimately failing.

My work on AUUGN and video stuff is converging: the former is now intended to come out on DVD, along with Fedora Core 4, and I need to work out how to handle them. A while back I used growisofs to burn images, but the name still annoys me, so today I dragged out some old mail messages describing how to use cdburn for this purpose:

Date: Thu, 15 May 2003 16:04:01 -0700
From: Joshua Oreman <oremanj@webserver.get-linux.org>
To: Jim Sills <jsills@g3llc.com>
Cc: questions@freebsd.org
Subject: Re: Writing DVD+RW

On Thu, May 15, 2003 at 02:45:39PM -0500 or thereabouts, Jim Sills seemed to write:
> Anybody riting DVD+RW disks??  Anybody ported growisofs??

As of 5.0 (and maybe 4.8), DVD+RW is supported in burncd.

To get you started (these commands all as root):
# Insert a DVD+RW
burncd -f /dev/acd0c format dvd+rw # <-- You only have to do this once
burncd -f /dev/acd0c -s max dvdrw imagefile.iso # <-- Burn a file
# The disk will be erased automatically if you burn something to
# it and there's already something on it
#
# Replace /dev/acd0c above with your burner /dev node
      

Well, it sort of worked. The format command made it to 78% before hanging, and on a repeat it was only 38%. An attempt to burn was even much less successful:

=== root@teevee (/dev/ttyp4) /spool/data 12 -> burncd -f /dev/acd0 -s max dvdrw FC4-i386-DVD.iso
next writeable LBA 0
writing from file FC4-i386-DVD.iso size 2686116 KB
written this track 1568 KB (0%) total 1568 KB
only wrote -1 of 32768 bytes: Device busy
      

Decided to try growisofs instead. That was better, but still not adequate. The session finished with:

builtin_dd: 1343248*2KB out @ average 2.3x1385KBps
/dev/pass0: flushing cache
/dev/pass0: stopping de-icing
/dev/pass0: writing lead-out
:-[ CLOSE SESSION failed with SK=2h/ASC=04h/ACQ=07h]: Resource temporarily unavailable
      

No idea what that message refers to; in any case, I was able to mount the image, but I found:

=== root@teevee (/dev/ttyp4) /spool/data 26 -> ls -l /cdrom/
ls: FC4-i386-DVD.iso: Value too large to be stored in data type
      

Further investigation showed that the image had been written to a Joliet file system as a single file, not quite what I was looking for. Presumably there's a 2 GB limit on file sizes, so the file system couldn't handle the size of this file.

Obviously I could go back and RTFM to find out how to do it correctly, but during discussions on IRC, Daniel O'Connor recommended k3b, which, it turns out, I had installed in March, though I didn't mention it in my diary. It's menu-driven, but I think burning CDs or DVDs is one case where that makes sense. What makes less sense is yet another example of breaking file system hierarchies:

 
This should be k3b-init.gif.  Is it missing?
Image title: k3b init
Dimensions: 790 x 546, 114 kB
Dimensions of original: 790 x 546, 114 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 19 June 2005:
thumbnails    small images    diary entry

This is yet another example of a small window (took up much less than half the height or width of the screen) with gratuitous scroll bars, and in this case, for some reason, unnecessarily large text (the images are reduced in size; click on them to see the original size). Not as if it could deal with text of that size:

 
This should be k3b-half.gif.  Is it missing?
Image title: k3b half
Dimensions: 546 x 205, 8 kB
Dimensions of original: 546 x 205, 8 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 19 June 2005:
thumbnails    small images    diary entry

After cleaning away the bottom and expanding the window, I got:


This should be k3b-menu.gif.  Is it missing?
Image title: k3b menu          Dimensions:          823 x 1109, 150 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Sunday, 19 June 2005, thumbnails          All images taken on Sunday, 19 June 2005, small
Diary entry for Sunday, 19 June 2005

 

What's Home? Root? They're inventions of k3b and bear no relationship to the file systems on the machine. The closest I could imagine are the directories /home and /root, both in the root file system. But they're not: other things were missing. By trial and error I found that Root is the root file system (/). But why has it been put on the same level as Home? And what is Home, anyway? I really had to go looking and comparing file names to work out what it is. It's not /home, which looks like this:

=== grog@teevee (/dev/ttypa) /spool/data 1 -> ls /home
Makefile        Sysconfig       Sysconfig.old   grog
      
It's not /home/grog either:
=== grog@teevee (/dev/ttypa) /spool/data 2 -> ls /home/grog
1.adv                                   Makefile.virgin
1024x576                                RCS
10ct:                                   README.LEMIS
1122                                    Rebekka
1122u                                   W
12-days-of-christmas                    X.README
12-days-of-christmas.c.commentary       XFree.oldmail
2a.adv                                  a.out
2andy                                   address.tex
2b.adv                                  calendar
      
Instead, it's /root:
=== grog@teevee (/dev/ttypa) /spool/data 3 -> ls /root
1024x576                axhome                  nsmail
Mail                    build-kernel            pr2send
Make.log                crontab                 rebuild
Makefile                dead.letter             root
News                    dobackups               sd0label
RCS                     emacs                   ssh
TMDA_DELIVERY_FAILURE   extract-updates         tmp
XEphem                  indent                  version
XF86Config.new          mail                    xorg.conf.new
XF86Config.new.older    merge-updates
XF86Config.older        mnt
      

That is unbelievably confusing. Two “folders”, one called Root which is really /, and the other called Home which is really /root, both put at the same level. Why do people do this? It makes it so much more difficult to understand. It's also worth mentioning that the current working directory (/spool/data, where I keep the ISO images) is missing from this view, which makes the whole thing relatively useless.

Finally got the thing running, and that at least works quite nicely, despite a particularly irritating habit the program has of automatically raising itself whenever the cursor is passed over it. At least the DVD was written correctly, though I was a little confused by the fact that at the end of the burning session I couldn't mount the DVD: it had been automatically ejected. But the DVD was OK, something that in the past has been a real issue with Fedora, so I suppose I can learn to live with the strangenesses.

Installing Fedora was another matter. I was using the old sat-gw machine, which had run Red Hat Linux for several years as my satellite downlink machine. It didn't install well, but there are enough reasons to believe that it's the hardware, even if it did run flawlessly for years. One way or another, it's nicer to be able to install from DVD and not have to constantly change CDs.


Monday, 20 June 2005 Echunga
Top of page
previous day
next day
last day

Came into the office this morning and looked at the machine I had been installing Fedora Core 4 on. It was showing the initial install menu. Who knows what happened there? I couldn't boot from disk, so did a reinstall (mercifully much faster; I wonder what it really did), but I still couldn't boot from it. I wonder if that's hardware or software.

We've changed our meeting schedule, and now we're having weekly meetings on Monday mornings. Into town in pouring rain—why has rainfall been such a binary thing this year—and had two meetings in succession, the second one mainly defining APIs, which looks like being my work for the foreseeable future.

I had to give back the Sony data projector at the meeting. I've ordered a new machine, a Panasonic PT-AE700E, but it hasn't arrived yet. Managed to get a loan of a PTLC-75, for which I can't find anything useful on the web; it's a 1024x768 unit, and the difference shows. Compared to the Sony (see my comments from 10 June),


Tuesday, 21 June 2005 Echunga
Top of page
previous day
next day
last day

API design today, none too easy. So much depends on what people expect, and each of us appears to expect something else.

In the afternoon to talk to Ramana Rao, an investment consultant. Looks like I'm now part owner of a vineyard in the Barossa.

The combination of the bad weather and the good ADSL service was too good to hold. This evening, by chance, I looked at the ADSL router statistics, and found that the upstream noise margin had diminished from about 23 dB to 6. Called Telstra and spoke to Julie, who of course told me I should take it to Internode, until I pointed out that it was a Telstra line (not to mention the fact that Telstra handles all the copper anyway). Finally she tested the line and got some less than perfect results, so at least I have a fault number (116269434). Hopefully this will turn out better than on previous occasions. Set up a cron job to monitor the line quality.


Wednesday, 22 June 2005 Echunga Images for 22 June 2005
Top of page
previous day
next day
last day

Up this morning to discover that the good news was that the ADSL line was still up. Spent some time investigating, including setting up a status monitor page. Things aren't looking too good, though:

 
This should be dslstats.png.  Is it missing?
Image title: dslstats
Dimensions: 540 x 360, 6 kB
Dimensions of original: 540 x 360, 6 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Wednesday, 22 June 2005:
thumbnails    small images    diary entry

In particular, the downstream margin frequently goes below 0, and there are service dropouts as a result. I may disappear off the net.

Writing date-dependent code is a pain. Here's a trick that uses FreeBSD's date program to convert from a file date (output in full with ls -lT) to seconds since the epoch:

  file=foobar           enter file name here
  date=`ls -lT $file | awk '{print $6 " " $7 " " $8 " " $9 }'`
  now=`date -j -f "%b %d %H:%M:%S %Y" "$date"  '+%s'`
      

Called “Call Australia” and spoke to somebody who didn't understand the issues, and who wanted me to disconnect the ADSL line so that he could test the (unrelated) phone line. I wish I understood the problems that people have in understanding that more than one phone line is involved. Spoke to Janice, his supervisor, who also took 5 minutes to understand that three different lines were involved. She still made me disconnect all lines except the one I was talking on, but then didn't do a test.

Later got a call back from Adrienne of Call Australia, who said that she would raise a fault and send me email. I didn't hear back from her.


Thursday, 23 June 2005 Echunga Images for 23 June 2005
Top of page
previous day
next day
last day

Got into the office at 8:30 this morning and discovered that all three phone lines were down. Strangely, the ADSL line wasn't, and the line stats hadn't deteriorated noticeably:

stats stats

Calls to both Telstra and Call Australia brought the information that they were working on “rehabilitating” the lines, and that they estimated to be complete at 1900 this evening.

In the afternoon, into town to a meeting of the ICT council of South Australia, and saw the linesmen on the way, not far from the location of the last fault. They confirmed that the line was between two other sections that had already been replaced. Why can't Telstra just replace the whole thing?


This should be Line-fault-1.jpeg.  Is it missing?
Image title: Line fault 1          Dimensions:          1728 x 2304, 1024 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Thursday, 23 June 2005, thumbnails          All images taken on Thursday, 23 June 2005, small
Diary entry for Thursday, 23 June 2005 Complete exposure details

 

They were waiting for a digger, and they were still confident to have things finished by this evening. On the way back, though, at 1730, there was no sign of the linesmen, no sign of digging, and no dial tone on the phone. So, once again, they have not honoured their Customer Service Guarantee. On the other hand, the DSL line (not covered by the guarantee) was still working, and since 8 am today the margins are better—maybe because of the lack of other traffic. I wonder if they work without DC voltage on the line.

I wonder how long it will be before I completely disappear from the Net.


Friday, 24 June 2005 Echunga Images for 24 June 2005
Top of page
previous day
next day
last day

First thing this morning, of course the phone lines were still down. Called Telstra, spoke to Becca, and was given a new repair date: 1900 tomorrow, Saturday 25 June. Put in a complaint, number 1-48335559, and wrote yet another letter to Alexander Downer; I doubt he cares.

At 9:45, even the ADSL line died. Prepared to go into town and do something else, but to my surprise it came back again at 11:10:

 
This should be dslstats.png.  Is it missing?
Image title: dslstats
Dimensions: 540 x 360, 4 kB
Dimensions of original: 540 x 360, 4 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 24 June 2005:
thumbnails    small images    diary entry

Obviously that was the time they needed to actually splice the cable. When it came back, the margin values were back to normal, but it only took 15 minutes for the downstream margin to drop by 5 dB. I wonder what caused that. Went and had a talk to the linesmen, but of course they didn't know either.

In the evening, by way of diversion, got a call from somebody who claimed to be from Telstra. She wanted to sell me ADSL. When I asked about the application I made in March (order number 0782927), she said “sorry, you'll have to talk to Telstra about that”. It's really borderline fraud for these resellers to say that they come from Telstra. I wonder when I'll finally hear something definitive about that order.

C't arrived today, along with a double-sided DVD with over 9 GB of data on it, including SuSE Linux 9.3. We (AUUG) are going to be steaming to get anything like that onto our DVDs. Spent some time trying to install it on the old sat-gw machine; it's still slow, but it seems to be a smoother install than Fedora Core 4.


Saturday, 25 June 2005 Echunga Images for 25 June 2005
Top of page
previous day
next day
last day

Spent some time today working on AUUGN, which is now due to come out on DVD. We seem to have plenty of space, and it looks as if we'll end up with Fedora Core 4, OpenSolaris, the Lions Book and the complete sources to the Sixth Edition of UNIX. The only problem is how to burn the DVD: it's basically supposed to be a modified Fedora DVD, but I don't know how to get it to boot. Didn't even make it that far with my attempts to burn with k3b, though; I take back those nice things I said about it last weekend. Today, for no apparent reason, it had modified its already tiny URL window to be exactly 7 overlarge characters wide. You'll need to select on this image to see anything even marginally useful:


This should be k3b.png.  Is it missing?
Image title: k3b          Dimensions:          2048 x 1536, 224 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Saturday, 25 June 2005, thumbnails          All images taken on Saturday, 25 June 2005, small
Diary entry for Saturday, 25 June 2005

 

There seems to be no way to influence text or window sizes. It also wasn't able to burn anything this time; whether due to incorrect usage or stupidity, I don't know, but the whole thing looks like a triumph of eye candy over functionality.

Installed SuSE 9.3 on eucla, my Dell Inspiron 9100 laptop. It's a surprising combination of good and not-so-good. The installation works well, though the error messages I got when setting up the network were very misleading: I had set up a total of three interfaces: Ethernet, wireless and modem. This was the first operating system that supports the modem, but it started trying to dial to connect to the Ethernet. It took a long time to discover that I had forgotten to set a default route for the Ethernet connection.

I can't help myself, I hate “modern” GUIs. SuSE comes with KDE, and it's painful. It's not helped by the unbelievably slow touch pad setup; I'll have to investigate that. It uses X.org, just like it does under FreeBSD, but getting the mouse to work properly is a real pain, and I had to manually convince it to use the real display resolution (1680x1200): it wanted to set up for 640x480 only. I didn't have any such problems when installing FreeBSD. SuSE also has strange programs that rebuild the configuration file, leaving just a message in it saying “don't edit this file”, but not telling you what you should edit. I can see myself using fvwm2 again. I wonder how much system functionality I'll lose as a result.

Replaced the fan in the air conditioner, which had been rewound. It doesn't work. And I was too stupid to test it before the painful work required to put it back in the machine. Grrr.


Sunday, 26 June 2005 Echunga
Top of page
previous day
next day
last day

More work on bootable DVDs today. Learnt a number of things about the bootstraps, notably that they're not like the first-level bootstrap on disks: they know something about the files on the DVD, so one size does not fit all. Tried with a script in the FreeBSD source tree, which built a “bootable” DVD, but the result was:

Building the boot loader arguments
Looking up /BOOT/LOADER...  File not found
Looking up /boot/loader...  File not found
Boot failed
      
Those files are right for FreeBSD, but not for Linux.

Next did some googling and found an article How to make a bootable Fedora ISO, which sounds pretty much like what I want. It refers to a directory isolinux, which really exists on the DVD, but the contents are:

=== root@teevee (/dev/ttyp6) /spool/data 113 -> l FC4-i386/isolinux/
total 8
-r--r--r--  1 root  wheel     2880 Jun  7 08:31 TRANS.TBL
-r--r--r--  1 root  wheel     2048 Jun  7 08:31 boot.cat
-rw-r--r--  1 root  wheel      292 Jun  7 08:09 boot.msg
-rw-r--r--  1 root  wheel     1034 Jun  7 08:09 general.msg
-rw-r--r--  2 root  wheel  4492627 Jun  7 08:09 initrd.img
-r--r--r--  1 root  wheel    10424 Jun  7 08:24 isolinux.bin
-r-xr-xr-x  1 root  wheel      595 Jun  7 08:09 isolinux.cfg
-r--r--r--  1 root  wheel    94600 Jun  7 08:09 memtest
-rw-r--r--  1 root  wheel      788 Jun  7 08:09 options.msg
-rw-r--r--  1 root  wheel      872 Jun  7 08:09 param.msg
-rw-r--r--  1 root  wheel      490 Jun  7 08:09 rescue.msg
-rw-r--r--  1 root  wheel      549 Jun  7 08:09 snake.msg
-rw-r--r--  1 root  wheel     5692 Jun  7 08:09 splash.lss
-rw-r--r--  2 root  wheel  1702587 Jun  7 08:09 vmlinuz
      

The problem is that isolinux.bin is the bootstrap, and for ISO 9660 images it should be a floppy image. It's obviously wrong. I tried padding it out to 1.4MB, but the result was that the system halted after loading it: clearly not the right bootstrap.

So where's it hidden? Conventional wisdom says that the bootstrap should be in the directory tree, and the burning software won't accept anything else, but I can't find any file on the disk which is exactly 1.44 or 2.88 MB in size. Maybe I'll have to extract it from the DVD with dd. There, at any rate, isoinfo helps:

=== root@teevee (/dev/ttyp6) /spool/data 119 -> isoinfo -d -i FC4-i386-DVD.iso
CD-ROM is in ISO 9660 format
System id: LINUX
Volume id: FC/4 i386 DVD #9
Volume set id:
Publisher id:
Data preparer id:
Application id: FC/4 i386
Copyright File id:
Abstract File id:
Bibliographic File id:
Volume set size is: 1
Volume set sequence number is: 1
Logical block size is: 2048
Volume size is: 1343058
El Torito VD version 1 found, boot catalog is in sector 270
Joliet with UCS level 3 found
Rock Ridge signatures version 1 found
Eltorito validation header:
    Hid 1
    Arch 0 (x86)
    ID ''
    Key 55 AA
    Eltorito defaultboot header:
        Bootid 88 (bootable)
        Boot media 0 (No Emulation Boot)
        Load segment 0
        Sys type 0
        Nsect 4
        Bootoff 10F 271
      
I suppose Bootoff is a sector number, but the format is strange.

Burning DVDs is still a gamble. burncd still fails:

=== root@teevee (/dev/ttyp5) /spool/data 10 -> burncd -f /dev/acd0 dvdrw AUUGN-26-1 fixate
next writeable LBA 0
writing from file AUUGN-26-1 size 3049120 KB
written this track 1568 KB (0%) total 1568 KB
only wrote -1 of 32768 bytes: Device busy
      

Once I build an ISO image, k3b will still burn it for me. While it was doing so, I took the opportunity to look at it with ps -ww. The command it used was:

/usr/local/bin/growisofs -Z /dev/cd0=AUUGN-26-1 -use-the-force-luke=notray -use-the-force-
luke=tty -speed=2.4
      

“-use-the-force-luke”? The man page just says

There're several undocumented options commonly denoted with -use-the-force-luke prefix. Some of them serve debugging purposes. Some require certain knowledge about recording process or even OS kernel internals and as being such can induce confusing behaviour. Some are to be used in very specific situations better recognized by front-ends or automated scripts. Rationale behind leaving these options undocumented is that those few users who would actually need to use them directly can as well consult the source code or obtain specific instructions elsewhere.

What a way to use software! In any case, heavily modified the FreeBSD burn script and came up with my own script which seems to do things without as much pain as “user friendly” programs like k3b.


Monday, 27 June 2005 Echunga
Top of page
previous day
next day
last day

Into town for the Monday meeting again, without anything of event happening. The meeting was shorter than normal today, but it still meant that I hardly got anything else done. Spent some time looking at the output of doxygen; some of it looks very good, but I need to come to terms with what it's telling me.

Got called out of the meeting by a phone call from Peter, a Telstra serviceman, who had turned up at our place to find out why there was noise on the voice phones. My best guess is that the people at Call Australia who took my complaint last Wednesday really had so little understanding that they thought the problem could be with the termination equipment. sigh.

Sent a message out to the Australian Linux groups today about my DVD boot problems, and in the evening got a reply from Glen Turner pointing to the problem: turned out that the bootstrap was correct after all, but that I had a bug in my burn script (now corrected): there are at least two kinds of bootstrap you can put in an ISO image. The older style is the El Torito bootstrap, effectively an image of a floppy. The other is a native bootstrap, which can be of any length. That's what I wanted, and you specify it with the -no-emul-boot option.

Not that that removed all problems: after that I could run the bootstrap, but it failed again:

ISOLINUX 3.08 2005-05-19 isolinux: Image checksum error, sorry...
      

More head-scratching to do there. I may actually have to install Fedora Core 4 to do it right.

Also in the evening, a phone call from another Telstra complaint manager. I had already had a voice mail message from Janet on Friday, and I was waiting for her to try again (should have done so by Saturday evening). Today, though, Karen called me; she appears to be higher up the scale, and was very helpful, apparently because of the help of Alexander Downer, to whom I had sent another letter on Friday. It seems that my doubts about his interest were overestimated. Not that they can help much now that the line is up again; the real proof will be what happens next time.


Tuesday, 28 June 2005 Echunga
Top of page
previous day
next day
last day

Into town for an ADUUG lunch today, at the Eros Ouzeri. The name suggested a cafe-like place, but in fact they have a cafe next door, and the Ouzeri is quite up-market. Also some of the best Greek food I've had in a long time.

Later in the afternoon to Playford capital for a farewell: the ICT council has been hosted there for nearly 5 years, but now we're moving out, and today was to mark the event. There was a certain melancholic atmosphere about the event.

Today also I finally got the new Panasonic PT-AE700E projector, and spent some time setting it up. It's not all roses: despite the specs, I couldn't get it to display normal a text mode display, and it requires a lot of padding either side of the image to display a complete image under X. I've frequently seen before that projectors do this (they cut off both left and right sides), but after using the Sony VPL-ES2, a cheap projector that handles just about anything you throw at it, I had thought it a thing of the past. With the PT-AE700E I had to modify the line timing significantly: at 1280 pixels per line (the default), the line total went from 1532 to 1660. In the process, with the help of xvidtune, I managed to get the display into such a state that I couldn't display it at all; the projector kept flashing it up for half a second and then turning off again for several seconds.

The projector wasn't the only thing that gave me problems: its native resolution is 1280x720, but the display card didn't handle that well, and added a lot of flimmering junk at the bottom, which I couldn't get rid of until I increased the line count to 736, so I finally ended up with the following mode line:

ModeLine    "1280x736"     99.00   1280 1392 1488 1660    736  737  740  776 +hsync +vsync
      
Presumably the corresponding “correct” mode line would be
ModeLine    "1280x720"     99.00   1280 1392 1488 1660    720  721  724  760 +hsync +vsync
      
I'll try that out when I have time to change the card.

Finally, mplayer appears to be Just Plain Broken when it comes to 16:9 displays. So many DVDs have 16:9 content, but there appears to be no way to display them on a 16:9 display with mplayer. There seem to be three possible settings: 16:9, 4:3 and 2.35. They don't work as expected: selecting 4:3 (the default) on a 16:9 display stretches the display horizontally to 16:9. Selecting 16:9 stretches it even further. In fact, I haven't found a way to get mplayer display correctly on a 16:9 display. The obvious display formats would be:

The only way I have found so far to get usable results is to get the projector to compress the 16:9 back to 4:3, thus counteracting mplayer breakage. Not exactly the reason I bought a wide screen projector.


Wednesday, 29 June 2005 Echunga
Top of page
previous day
next day
last day

Spent most of the day investigating code in preparation for a clear API definition. In the process went through a number of source code analysis tools. My usual tool is etags, but that doesn't really help show call graphs, so looked at our doxygen documentation as well. That does fine in a downward direction, but in my case I'm trying to fit an API under the package, so I really need to see what is calling me, and doxygen doesn't do that. Installed Source Navigator, which Jeffrey Hsu introduced me to last year, but it turns out it also only traces down. Somebody mentioned kscope, a name which scared me, with reason: it's a KDE package, and just installing it pulled in 86 (really!) other packages, including Samba and OpenLDAP. Once I started it, it produced no useful output, and though I haven't given up, the interface turns me off: yet another package that doesn't understand the concept of working directories, and which gives me popups with windows too small to enter text in. It's reminiscent of k3b, with which I had so many problems in the last week or two. Didn't end up doing much source code analysis, but it's becoming relatively clear that the tools available aren't worth the trouble.

In the evening, more investigation of mplayer on wide screens, and discovered that there is, indeed, a command line option to tell mplayer what it should already be able to find out from the monitor dimensions: -monitoraspect 16:9 does it. It doesn't help display wide screen images on wide screens, though: “fullscreen” really means “full height”, not “full width”, and I wasn't able to find a way to change that. The only other options are “normal size” and “double size”; the latter was too big. Looks like some code work is needed.


Thursday, 30 June 2005 Echunga Images for 30 June 2005
Top of page
previous day
next day
last day

More work investigating source code tools today, in the middle of which Geoff Nairn came along to pick up some data for our test corpus. In the process, I showed him where I was with kscope, and we actually got it to work and import some files. It's still difficult to use, though I was able to reset some, but not all, of these enormous fonts.

Then Geoff left, which involved breaking into his car. He had locked the keys inside, but the security of the locks wasn't too hot, and it only took us 10 minutes to break in, most of it devising a method. After that, looked again at kscope, and indeed it can display an upside-down calling function tree:

 
This should be calltree.gif.  Is it missing?
Image title: calltree
Dimensions: 527 x 466, 18 kB
Dimensions of original: 527 x 466, 18 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 30 June 2005:
thumbnails    small images    diary entry

Once again I have this terrible layout; I don't know how to get the boxes at the bottom left to display the complete text. I've already told it to use smaller fonts, which it does in some places, but not here.

This is the window thrown up after you manually enter the name of the function (bp_wrap_file_write) and click on all the + boxes up to the top level call (in this case, process_connection). But I'm left wondering if it's worth the trouble:

In fact, the window contains (rather than displays) more information: it's hidden behind the scroll bars:

 
This should be calltree-2.gif.  Is it missing?
Image title: calltree 2
Dimensions: 526 x 452, 18 kB
Dimensions of original: 526 x 452, 18 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 30 June 2005:
thumbnails    small images    diary entry

You still can't do anything with it; the corresponding cscope output is more useful, since despite its unfriendly interface, it at least allows you at the sources.

I've never liked eye candy and GUIs for the sake of it, but in this case a GUI would be a good idea. It's sad that it has failed so badly. Sent out a message to the FreeBSD mailing lists to see if there's something else worth using, but I'm gradually coming to the conclusion that this is a waste of time.

In the evening took another look at mplayer with an aim to getting it to handle 16:9 better. What horrible looking code! It has no comments, non-uniform indentation and spacing that suggests that it's really preprocessor output. Spent some time tidying up the code—I can see that I'm in the process of a minor fork here—and got rid of at least one irritating error message. Much more to be done.


Top of page Previous month Greg's home page Today's diary entry Next month Greg's photos Copyright information

Valid XHTML 1.0!

$Id: diary-jun2005.php,v 1.55 2014/01/28 00:17:57 grog Exp $