|
| Greg's diary |
| January 2008 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Tuesday, 1 January 2008 | Dereel | Images for 1 January 2008 |
| Top of page | ||
| next day | ||
| last day |
And into another new year! How time flies! As if to reinforce the thought, my daily cron job told me;
80% life: Saturday, 24 February 1996 Feb 24 "Installing and running FreeBSD" submitted for publication, 1996
The “80% life” refers to my own lifetime; in other words, since submitting the book for publication, I've grown 25% older.
It seems that I'm not the only old grump around; J. D. Frazer, better known as Illiad, the cartoonist of “User friendly”, today published the following cartoon:
Unfortunately the “User Friendly” site has gone away. I'm still looking for a replacement link.
Technology may have improved, but somehow the average intelligence of computer users is dropping.
Spent some time playing around with Google Maps, something that usually annoys me. This time I was working out my itinerary from Melbourne to Dereel, but of course first tried their recommendation, which of course took me to the junction of the Dereel—Rokewood Junction Road and Judge Road. After adjusting it to what it thinks my address is, discovered that the trip from Melbourne to Dereel went via Ballarat, and the trip from Dereel to Melbourne went via via Geelong. Certainly it lends credence to my suggestion that we're evenly balanced.
Also discovered how to “save” custom maps; there's no button for it, but there is a link “link to this map”, which effectively saves it and generates a URL about 300 characters long; an excellent use for tinyurl.com. Choosing an alternative path isn't the simplest, especially on freeways. Firstly the software (Javascript? I'm sure somebody will correct me) is terribly slow, probably not helped by the satellite connections, and secondly it's difficult to position exactly.
For example, the directions from Lara to Werribee are pretty straightforward, but if you move one point on the freeway entrance back a few metres—something that can happen easily when you're changing a route—it'll take you up to the next exit, turn you round and take you back to the previous exit, turn you around again to take you past that particular point. It's understandable, but quite confusing, and the slowness of the software makes it really tedious.
| Wednesday, 2 January 2008 | Dereel | Images for 2 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Our hebes are gradually dying—another one died yesterday—so now that the weather isn't quite as hot, decided to plant the rest, a total of 29 confirmed cases and three other bushes which we had been told were hebes, but about which we have our doubts. Hopefully they'll be happier now. And only another 40 odd native trees and shrubs to plant.
In the last few weeks the bird activity has increased significantly. I've known from earlier that there are some very large flocks of galahs and sulphur-crested cockatoos, but lately the cockatoos have been very active, flying round the house and making an incredible din:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It's been nearly two weeks since my satellite connection got installed, and things still aren't working right. Finally got a call from Phil to tell me that IPStar are planning to trace my connection, and that I should “open a command line” (and look inside?) on Microsoft “Windows” and run NETSTAT; I'd never heard of this Microsoft program before, but what I saw suggests that it's a pretty faithful, if basic, copy of BSD netstat. It eventuated that he wanted to know how many established TCP connections I had. I told him that I had supplied this information long ago, and that the number gradually decreased to zero. Tried again, with “NettGain” (i.e. BST, as far as I can tell) enabled, and the same picture occurred: nothing has changed in two weeks, with the possible exception that somebody might finally be listening. After it hung up after less then 3 hours, tried rebooting with BST disabled; the first time it rebooted and automatically re-enabled BST, and I had to reboot for a third time for the setting to hold.
In the process, a listen on port 222 reappeared. As David Adam suggested, tried running nmap -sV and discovered:
=== root@eureka (/dev/ttypd) ~ 319 -> nmap -sV sat-gw
Starting Nmap 4.11 ( http://www.insecure.org/nmap/ ) at 2008-01-02 13:41 EST
Interesting ports on 192.168.5.100:
Not shown: 1672 closed ports
PORT STATE SERVICE VERSION
135/tcp filtered msrpc
136/tcp filtered profile
137/tcp filtered netbios-ns
138/tcp filtered netbios-dgm
139/tcp filtered netbios-ssn
222/tcp open tcpwrapped
445/tcp filtered microsoft-ds
8080/tcp open http Boa HTTPd 0.94.13
MAC Address: 00:07:93:43:22:14 (Shin Satellite Public Company Limited)
Nmap finished: 1 IP address (1 host up) scanned in 8.373 seconds
Interesting stuff, though I'm not sure how much more that helps me with accessing port 222. But port 8080 (http proxy acting as http) was interesting: Boa is a GPL program, but nobody offered me the sources. So my course of action is clear: first, get the link working, then go and “ask” for the source code for the web server. I'll be interested to see what happens.
| Thursday, 3 January 2008 | Dereel → Melbourne → Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Up early today to go shopping in Melbourne. Not an ideal day; the temperatures in Melbourne were predicted to be 30°, and it seemed like it. To the Footscray and Queen Victoria Markets, where we spent a lot of money in very short time and returned quite tired.
My satellite connection, without BST had stayed up since yesterday evening. Was just writing up the latest state of play, round 16:00, when I got a call from Peter to tell me that they had exhausted all options, and that they would replace the satellite modem soon. I told him about the experiences without BST, and he told me that IPStar had just turned it on again. I find it quite objectionable that people who aren't even prepared to talk to me go and mess around with my configuration. In any case, they proved the point: I now have a script to restart my SMTP tunnel, and it shows:
Read from remote host ozlabs.org: Connection reset by peer
Wed Jan 2 18:47:23 EST 2008
Read from remote host ozlabs.org: Connection reset by peer
Thu Jan 3 20:33:06 EST 2008
Read from remote host ozlabs.org: Connection reset by peer
Thu Jan 3 20:56:43 EST 2008
...
That continued through the night, though at least it was possible to reestablish the connection.
Also a call from a rather confused Phil, who apparently hadn't been informed of what was going on.
| Friday, 4 January 2008 | Dereel | Images for 4 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
How much work is there in preparing for a barbecue? You go down to the supermarket, buy a barbecue meat pack and some beer and wine, bring it home, throw the meat on the barbie, and you're done. But somehow that doesn't work for us. Admittedly, part of the work involved tidying up the place so that it was presentable, and moving one of the old conference tables from the Kreditanstalt für Wiederaufbau under the trees, where we'll have the meal, weather and insects permitting. Also a lot of food preparation. Made not one, but two different recipes for barbecued chicken wings; we'll see tomorrow how they compare.
We'll need networking outside, of course, so put in my second 802.11g access point. Had a lot of difficulty setting it up, and finally concluded that it's defective. So I had to move the other one closer, involving even more cables over the floor. sigh
Even Lilac got in on the preparations, though I don't suppose the meat she brought was intended for the barbecue:
|
|||||||||||||||||||||||
The rabbit was unharmed. She's done this once before, and we found her in her basket on top of the fridge with the terrified baby rabbit; I was too slow off the mark to take photos on that occasion. It's possible that she would not have killed this one either, but on another occasion the rabbit she brought in no longer had a head, so saved the rabbit and let it out, much to Lilac's disgust.
My mutilated bottlebrush is coming back strongly, with lots of new shoots. One of the problems we had had was that the bottlebrush blooms, the flower dies, but the new growth goes through the dead flower. Today I confirmed that most of the new shoots came from behind the dead flowers, so the last one of them is gone. It looks as if it's going to do well, though: I think we might even be getting more flowers.
| Saturday, 5 January 2008 | Dereel | Images for 5 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Lots of preparation work today for the barbecue, then over to the Yeardleys for the marriage ceremony between David and Fifi, at which Chris and I were the witnesses:
|
|||||||||||||||||||||||
Back home and had a long wait for the BUGS contingent (Edwin Groothuis, Sam Lawrance and Juha Kupiainen), whom we finally located in Ballarat round the time the first steaks were ready. They showed up after a while and we had a pleasant time, despite the heat (36°). Despite the incredible quantities of food we had bought, we nearly got through it all.
|
|||||||||||||||||||||||
We didn't do much hacking this year, however; possibly the heat was a contributing factor. The BUGS people stayed the night, so we carried on into the evening, though the others had been up since 4 am to get the flights down here, so they didn't stay up late.
| Sunday, 6 January 2008 | Dereel | Images for 6 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
A different breakfast today: Edwin cooked some Poffertjes, which I've never had before.
|
|||||||||||||||||||||||
Not bad, but too sweet for me. Then they headed off along the Great Ocean Road and left me to process the photos I took yesterday. To my surprise there were 800 MB of them. That took up much of the day.
The bottlebrush that I pruned so radically last month is coming back surprisingly strongly. The first photos was taken on 23 December, and the second one today; in the meantime I've pruned away the stuff in the background, but it's clear how much things have grown in that time.
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Coincidentally watched an article on ABC's Gardening Australia programme which confirmed that this was the correct way to prune bottlebrushes.
| Monday, 7 January 2008 | Dereel | Images for 7 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Pruning bottlebrushes
|
Topic: gardening | Link here |
And suddenly there's nothing pressing to do! Spent most of the day in the garden, mainly pruning trees. I'm gradually getting to understand bottlebrushes better. The problem is that when the flowers die, they don't fall off:
|
|||||||||||||||||||||||
The spheres with holes in them are the remains of individual stamens. If you don't do anything, new growth occurs on the other side. This one shows a stem grown through the remains of two years' flowers:
|
|||||||||||||||||||||||
I've found that there are shoots immediately behind the flowers, so I've been cutting them off very short. Spent some time pruning all this stuff away, with noticeable results:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
TCP connectivity
|
Topic: technology | Link here |
Also wrote up a script to test the TCP connectivity of my satellite connection. While doing so, got a call from Daniel at Wideband. While discussing things, the connectivity completely went to hell, in a way that I haven't seen before: existing connections hung, while I was still able to establish new ones, just the opposite of what I've seen so far. It went away when I turned BST off again. I wonder how much longer this is going to go on.
| Tuesday, 8 January 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More work on my TCP speed graphs today, and got them to a point where they could show something. They immediately showed some interesting things:
The most obvious thing here is the improvement in general connectivity yesterday round 21:00. This was probably IPStar fixing the presumably more general problem that complicated problem investigation yesterday. It also showed a relatively steady fetch time of about 4 seconds (speed 0.25), up to about 14:00. More of that below.
According to the broken web server in the modem, NettGain was enabled. They seem to use the terms BST and NettGain interchangeably. I'll stick to BST; the spelling “NettGain” annoys me. So what effect does it have? Did some more research and read a white paper that wasn't very specific, though it does show what I had expected and shows a diagram on page 6 that shows BST bridging the TCP connection between the satellite modem and the other end of the satellite link. It doesn't go into specifics about the protocol, but it's reasonable to assume that the first SYN packet should get a local reply, like many proxy servers do. That again should be noticeable from the time it takes to arrive. Here's what I saw after fixing tcpdump's horribly broken formatting of time intervals (first field on each line, either just microseconds with leading 0s, or seconds and fractions):
000000 IP 192.168.5.101.55869 > ozlabs.org.http: S 3339708160:3339708160(0) win 65535 < mss 1460,nop,wscale 1,nop,nop,timestamp 1313538167 0,sackOK,eol> 1.420659 IP ozlabs.org.http > 192.168.5.101.55869: S 2906788800:2906788800(0) ack 33397081 61 win 5792 <mss 1460,sackOK,timestamp 880734521 1313538167,nop,wscale 6>
Here we have eucla (192.168.5.101 on this interface) sending out a SYN packet and getting a SYN, ACK response 1.42 seconds later. With the still to be proven hypothesis that the software in the modem isn't completely broken, this suggests that the packet went to the other end of the link and back again. The rest of the transfer continues. eucla ACKs the SYN, ACK packet and sends out an HTTP request GET /grog/emptyfile HTTP/1.1:
000547 IP 192.168.5.101.55869 > ozlabs.org.http: . ack 1 win 33304 <nop,nop,timestamp 1313 539589 880734521> 000781 IP 192.168.5.101.55869 > ozlabs.org.http: P 1:119(118) ack 1 win 33304 <nop,nop,tim estamp 1313539589 880734521>ozlabs responds 2.45 seconds later, quite a time even for a satellite connection. But things don't always seem to be that bad.
2.456031 IP ozlabs.org.http > 192.168.5.101.55869: . ack 119 win 91 <nop,nop,timestamp 880 735138 1313539589> 002574 IP ozlabs.org.http > 192.168.5.101.55869: P 1:360(359) ack 119 win 91 <nop,nop,time stamp 880735139 1313539589> 000068 IP ozlabs.org.http > 192.168.5.101.55869: F 360:360(0) ack 119 win 91 <nop,nop,time stamp 880735139 1313539589>The last packet has the FIN flag set, so eucla shuts down the connection:
000055 IP 192.168.5.101.55869 > ozlabs.org.http: . ack 361 win 33304 <nop,nop,timestamp 13 13542049 880735139> 001286 IP 192.168.5.101.55869 > ozlabs.org.http: F 119:119(0) ack 361 win 33304 <nop,nop,t imestamp 1313542050 880735139>ozlabs responds with the final ACK:
1.047480 IP ozlabs.org.http > 192.168.5.101.55869: . ack 120 win 91 <nop,nop,timestamp 880 735368 1313542050>
So the total time for this transfer was about 5 seconds. The graphs show that 4 seconds is more typical, involving three RTTs, of which only the second is payload. So what is BST doing here? As far as I can tell, nothing. It's not clear that it's even enabled; the web application appears to say that it is, but can I trust that? Rebooted and discovered that no, I can't trust it, and yes, indeed, things change—not always for the better:
000000 IP 192.168.5.101.60482 > ozlabs.org.http: S 2067612446:2067612446(0) win 65535 <mss 1460,nop,wscale 1,nop,nop,timestamp 1398268529 0,sackOK,eol> 000998 IP ozlabs.org.http > 192.168.5.101.60482: S 585494650:585494650(0) ack 2067612447 w in 8192 <mss 1460>
Here the ACK comes in 998 µs, which clearly can't have come from the other end of the link.
000138 IP 192.168.5.101.60482 > ozlabs.org.http: . ack 1 win 65535 000168 IP 192.168.5.101.60482 > ozlabs.org.http: P 1:119(118) ack 1 win 65535 000770 IP ozlabs.org.http > 192.168.5.101.60482: . ack 119 win 8192 11.546367 IP ozlabs.org.http > 192.168.5.101.60482: P 1:360(359) ack 119 win 8192 000482 IP ozlabs.org.http > 192.168.5.101.60482: F 360:360(0) ack 119 win 8192
Unfortunately, for reasons I don't understand, it now takes 11.5 seconds to actually get the file, much longer than the whole transaction without BST.
000103 IP 192.168.5.101.60482 > ozlabs.org.http: . ack 361 win 65535 001403 IP 192.168.5.101.60482 > ozlabs.org.http: F 119:119(0) ack 361 win 65535 000689 IP ozlabs.org.http > 192.168.5.101.60482: . ack 120 win 8192
Again, eucla sends a FIN, ACK sequence and gets the final ACK in 689 µs. So BST is working—or is it? Looking at the graphs, the smoothed graph shows that the average speed is a little better better, but there's much more deviation. This appears to be repeatable, as the raw log files show:
1199843909 2.08 # Wed Jan 9 12:58:31 EST 2008 1199843975 1.94 # Wed Jan 9 12:59:37 EST 2008 1199844042 7.33 # Wed Jan 9 13:00:49 EST 2008 1199844060 11.44 # Wed Jan 9 13:01:12 EST 2008 1199844078 2.28 # Wed Jan 9 13:01:21 EST 2008 1199844096 11.66 # Wed Jan 9 13:01:48 EST 2008 1199844116 2.56 # Wed Jan 9 13:01:58 EST 2008 1199844183 11.58 # Wed Jan 9 13:03:15 EST 2008 1199844202 2.43 # Wed Jan 9 13:03:24 EST 2008 1199844269 11.55 # Wed Jan 9 13:04:40 EST 2008 1199844287 11.23 # Wed Jan 9 13:04:58 EST 2008 1199844356 2.02 # Wed Jan 9 13:05:58 EST 2008 1199844423 11.45 # Wed Jan 9 13:07:14 EST 2008 1199844442 2.15 # Wed Jan 9 13:07:24 EST 2008 1199844459 11.45 # Wed Jan 9 13:07:51 EST 2008
The first field is the timestamp in seconds (decoded in the comment on the right), and the second is the time in seconds that it takes for the total transfer. It's interesting that nearly all of them take either 2.5 seconds or less, or over 11 seconds (matching the example above), with very little in between. Before the connection dropped completely round 17:30, there were a couple of successful transfers, but not many. And finally, at about 18:30, I rebooted the modem and we were back to the start again.
So where does this leave me? Clearly BST isn't fulfilling its promise. It now seems clear that it's at least partially responsible for the drop of TCP connectivity, but the new recognition is this strong fluctuation in TCP transfer times. Does this mean that the modem is defective, or that BST is misconfigured for my connection? Again, something that only IPStar can analyse. The fact that they haven't done so in 3 weeks says a lot about the quality of their service.
| Wednesday, 9 January 2008 | Dereel | Images for 9 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
We've been here 6 months! Things have definitely changed, although we can't claim to have completed the move in yet—there are cables all over the floors, and the dining room is still full of tools:
|
|||||||||||||||||||||||
Still, it feels like home now, and Wantadilla is becoming a memory.
Into town for a blood test and to look for some irrigation equipment, the latter unsuccessfully. Stopped in at VicRoads to find out what I had to do to transfer my cars to Victorian plates, and discovered it wasn't much: left 30 minutes later with the Magna on its third set of plates (VGM 540 was stolen five years ago, and now WXD 672 had been changed to WFP 292 (“Victoria: The place to be because the speed limits are so slow that you can't go anywhere else”)).
Then to have a haircut and home. Spent the afternoon messing around with more TCP monitoring stuff.
|
No road signs in Ballarat?
|
Topic: general, opinion | Link here |
After coming here, Edwin Groothuis wrote in his blog:
If you think that (insert a random town) is badly sign-posted, Ballarat is worse. On the eight kilometer trip to the information center we found four sign-posts with an "I" on it. And one indication said "<- B300 ->" but it didn't say which places it went too.
Clearly Edwin hasn't been to Adelaide. Certainly road signs could be improved, but I think he's being unfair to Ballarat. He's also being inaccurate: it's unlikely that any sign in Ballarat says “B300”, which runs between Benalla and Barajarg, nearly 300 km away. Probably Edwin meant the Midland Highway, the A300. And it has plenty of signposts. Here are the ones they would have driven past on the way to my place.
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Edwin continues:
Sign posts in Victoria don't have distances on it. They tell you which city/town/county/township you are going to, and which road number you are on, but it doesn't tell how far away you are. Except for Ballaret [sic], they just give you the road number.
We've seen a counterexample above. Even on the C roads, there are plenty of distances:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Finally we also see an example of a sign with just the road number and arrows. These aren't supposed to tell you where to go, of course, just where you have got to. This example is in a small road in Napoleons. And, of course, it is accompanied by more detailed directions. I know that there are signs like this on the A300, but I didn't see them during my search.
Finally, Edwin writes:
Gregs house is on the right hand side of the road, while his mailbox is on the left hand side of the road.
Not really. The house is on the left, and the mail box is on the right. That's because the postman comes out of the swamp and drives on, so all mail boxes have to be on the same side of the road. Clearly the BUGS people missed this turnoff:
|
|||||||||||||||||||||||
That's easy to do; they would then have gone past the swamp, like the postman does, though it's still difficult to understand how they missed the turnoff when I had given them this map.
So what is Edwin really saying about the signs in the Ballarat area? I can think of some less obvious things:
Clearly some form of standardization is a good idea. Things are happening, of course, but probably not on the scale that is needed. The photos above show two different styles of signpost, the old, small, black and white ones and the newer green, white and yellow ones. The latter are easier to recognize, but they bear little relationship to the signs in Europe, for example.
| Thursday, 10 January 2008 | Dereel | Images for 10 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Today I received another letter from Telstra's Customer Relations department, who seem to operate completely independently from the rest of the company. They gave a reference which apparently was supposed to be a complaint reference number, but one I had never seen before, and told me it had been closed. How? Why? So far Telstra has not resolved a single problem that I have had. Wrote a reply, also sent to my MP. I have little hope that they will resolve anything.
The weather is living up to its promise of a long, hot summer. Today it reached 40° and stayed at that temperature for much of the day. Despite watering the garden for half the day, many plants showed signs of stress; one of the newly planted trees may die. It's fairly clear that, for the moment, planting the remaining trees is out of the question. Hopefully they'll survive like that until autumn.
| Friday, 11 January 2008 | Dereel | Images for 11 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The weather was supposed to suddenly get cooler today, but it didn't quite happen like that. At 8:30 am it was 30°, and it rose to 36° by mid-morning before dropping slowly for the rest of the day, reaching 22° by 6 pm. Strange weather.
Plenty of computer frustration today. I've not been very happy with GANDI, the domain registrar, in recent times, mainly due to their absolutely impossible web site. I've been meaning to change to a private registrar, but it seems that I've left it too late. Went to the web site and tried to update it. That in itself wouldn't have been too difficult, but the “pro-forma” invoice (what's pro-forma about an invoice that you have to pay?) showed the wrong address.
How do you change the address? Changing my own (via my GANDI handle) works, for a certain definition of “works”: they turned my address around so that the zip code came before the town name, which works in France, but not here. But I found absolutely no way to change my company details, which even last time I couldn't enter correctly: they wanted a state name, but the Australian states weren't on the list, so I ended up with the address:
owner-address: PO Box 460 owner-address: 5153 owner-address: Echunga owner-address: Andhra Pradesh owner-address: Australia
Today I found no way to change it. Finally gave up and went back to renewing the domains. This time I didn't get a pro-forma invoice, and there was no way to confirm whether my address details were correct or not.
Gave up trying to find out and went on to payment, via VISA. That took me, via Crédit Mutuel, to a Verified by VISA page, where I had to type in my details and password. Nothing new; I had registered with them before and had the details handy. Did that, pressed Submit—and nothing happened.
The page did specify that it wanted some strange plugin, and offered to load it for me. Why does firefox assume you're running Microsoft? I've never had any success with such downloads, so tried again on boskoop with Apple's safari. It didn't know the plugin either, and strangely didn't even offer to download it. Finally called the ANZ help line, fought my way through the voice menus to Verified by VISA support, and spoke to Simon, who after asking me what product I was using (why did I have to fight the voice menu?) told me that I wasn't registered for Verified by VISA. He also told me “It works with [Microsoft] Internet Explorer” and suggested that I should try to find a way to download the plugin manually. I told him that I didn't trust unknown plugins, and that any dependency on them was a serious security risk, but of course he didn't understand. Finally he suggested I tried to register again, which I did—and found that they had changed the password rules. They're stupid enough anyway, requiring only letters and digits, but at least one of each. My previous password, a comment on the stupidity of this scheme, was 7 characters long, but now they've introduced a minimum length of 8 characters, so my password was invalid. Instead of producing an error message, though, it just did nothing. If people can't even program a simple web application, would you trust a plugin that they provide?
Spent some time updating my map of Google Maps breakage round Dereel. In the process, a popup told me that I had no Google Profile. What's that? I do have a gmail account, but there was no mention of profiles there. Finally did what people do when looking for things: I used Google and found a help page that told me what it was about, but not how to set one up again. Tried another search, which brought me about 73,200 results. None of the 100 on the first page were from Google. Chose the first one, which did, indeed, tell me how to set it up, using tiny, non-resizable windows that once again seem to require tiny screens to work correctly; the submit button was not displayed, and there were no scroll bars, though with a couple of presses of the TAB key I was able to find them. So now I have a Google profile. What use is it? I have no idea.
The satellite connection seems more reliable at the moment. Am I really the only person who is having these problems with IPSTAR? One way to find out would be to run my monitoring software on other systems. Chris Yeardley's an obvious guinea pig, so set to adapting it to run on Mac OS X. That involved installing gnuplot, of course. While trying to do that, it appeared that the satellite connection had dropped after all, but while I was looking at that, eureka panicked with a general protection fault. When it came back up again, for some reason natd wasn't running, and even when I started it, I had no connectivity.
Finally fired up “Internet Explorer” on pain and found:
|
|
||||||||
That repeated after a couple of power cycles, during which “Internet Explorer” hung itself up and required the system to be rebooted, so tried to call Wideband—but the phone links from MyNetFone were overloaded, and I first had to rewire the connections to use POTS instead of VoIP. Finally got through and spoke to Brendan, who said he'd check and call back. He did so pretty soon to say that, in preparation for the change of modem on Monday, IPStar had disabled my login. It had been reinstated, but what kind of company does things like that? If they have to disable it, it should be done in conjunction with the technician who installs the hardware. It looks as if I'm going to be disconnected on Monday again. As it was, I had over an hour of down time:
|
||||||||||||||||||||||
| Saturday, 12 January 2008 | Dereel | Images for 12 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More work on the satellite statistics scripts today, and put together a package for download. I'd be interested to see how other IPStar users fare. Over to the Yeardleys and installed it on delicious, which went relatively smoothly and showed that yes, indeed, Chris was getting much better performance than I am. Still, a good data point. Back home and installed on boskoop, which is almost identical to delicious, and ran into all sorts of problems getting it to run under cron, with strange messages about corrupted scripts and things. Didn't get it sorted out by evening.
| Sunday, 13 January 2008 | Dereel | Images for 13 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Finally it's cooled down enough to go riding, so did that. Having the forest right outside the house has its advantages: since we've been here, we haven't used our horse float at all, and we've let the registration lapse as a result. But the forest isn't very big, and today I thought we could go out in the other direction. Took a look at Google Maps, and after establishing which roads really exist and which are just a figment of Google Maps' imagination, set off along what Google Maps calls Bliss Rd. It was pretty uninteresting, and soon we set off back into the forest to the south. Enfield State Forest is only barely to the north of us, but it's far enough to want to use the float, so I suppose we will do that soon.
| Monday, 14 January 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Today was the day that my satellite modem was to be changed—finally. John showed up with an old-style modem, in itself probably not a bad idea, but it came without the necessary stand, and when I looked in the box for the instructions, all I found was an RMA authorization dated 30 April 2007 describing power supply faults. Clearly it's been repaired since then, but I'm not very impressed with being fobbed off with a reconditioned unit with missing accessories.
On the plus side, it did seem to solve the BST problems: the TCP speed has now increased from 0.25 to 0.5. Instead, it introduced a new one: every time there's a glitch in the satellite link (far too often; red line in the following graphs) I get an ECONNRESET error (“Connection reset by peer”):
This appears to be one of the downsides (misfeatures?) of the BST protocol, and may make it useless unless the satellite connection is improved.
While tracing some stragglers, accidentally copied a version of this diary with some raw tcpdump output to my external web site. Peter Jeremy picked on it and asked:
<peter> grOogle: Is there a reason why today's diary entry is just a mangled tcpdump output?
There's a reason alright: that's the way it is. It's not just mangled, it's also confusing and downright ugly. I had already referred to it on 8 January: for some reason, the time format with the -ttt flag contains a spurious space if the time is a second or more. Here's an example of the output I was looking at; it was a single TCP transfer which, owing to link problems, involved retries:
009599 IP 192.168.5.101.62958 > ozlabs.org.http: S 742950541:742950541(0) win 65535 <mss 1460,nop,wscale 1,nop,nop,timestamp 1830705845 0,sackOK,eol>
2.990222 IP 192.168.5.101.63529 > ozlabs.org.ssh: S 488474110:488474110(0) win 65535 <mss 1460,nop,wscale 1,nop,nop,timestamp 1830708835 0,sackOK,eol>
009920 IP 192.168.5.101.62958 > ozlabs.org.http: S 742950541:742950541(0) win 65535 <mss 1460,nop,wscale 1,nop,nop,timestamp 1830708845 0,sackOK,eol>3.192179 IP 192.168.5.101.63529 > ozlabs.org.ssh: S 488474110:488474110(0) win 65535 <mss 1460,nop,wscale 1,nop,nop,timestamp 1830712035 0,sackOK,eol>
008881 IP 192.168.5.101.62958 > ozlabs.org.http: S 742950541:742950541(0) win 65535 <mss 1460,nop,wscale 1,nop,nop,timestamp 1830712045 0,sackOK,eol>3.189862 IP 192.168.5.101.63529 > ozlabs.org.ssh: S 488474110:488474110(0) win 65535 <mss 1460,sackOK,eol>
009988 IP 192.168.5.101.62958 > ozlabs.org.http: S 742950541:742950541(0) win 65535
<mss 1460,sackOK,eol>3.192507 IP 192.168.5.101.63529 > ozlabs.org.ssh: S 488474110:488474110(0) win 65535 <mss 1460,sackOK,eol>
008334 IP 192.168.5.101.62958 > ozlabs.org.http: S 742950541:742950541(0) win 65535 <mss 1460,sackOK,eol>
The connection that interests us is in bold; the grey text is a different connection, but I can't remove it without messing up the times. In this case, we didn't get a response from ozlabs.org at all. But the times (first and possibly second fields) look terrible! They always include 6 digits microseconds with leading 0s, and if the time is over a second, the second part is prepended—with a space. This is obviously a bug, and as Peter points out, makes processing with awk or similar programs almost impossible.
So: we have the source. Why not improve on it? Set to and did just that. But how? There are two possible ways, suitable for different things. Here's the first:
0.000000 IP 192.168.5.101.50355 > ozlabs.org.http: S 2624072652:2624072652(0) win 65535 <mss 1460,nop,wscale 1,nop,nop,timestamp 326824064 0,sackOK,eol>
0.001736 IP ozlabs.org.http > 192.168.5.101.50355: S 363258612:363258612(0) ack 2624072653 win 8192 <mss 1460>
0.000070 IP 192.168.5.101.50355 > ozlabs.org.http: . ack 1 win 65535
0.000119 IP 192.168.5.101.50355 > ozlabs.org.http: P 1:119(118) ack 1 win 65535
0.001957 IP ozlabs.org.http > 192.168.5.101.50355: . ack 119 win 8192
2.177414 IP ozlabs.org.http > 192.168.5.101.50355: P 1:360(359) ack 119 win 8192
0.000573 IP ozlabs.org.http > 192.168.5.101.50355: F 360:360(0) ack 119 win 8192
0.000064 IP 192.168.5.101.50355 > ozlabs.org.http: . ack 361 win 65535
0.000942 IP 192.168.5.101.50355 > ozlabs.org.http: F 119:119(0) ack 361 win 65535
0.002176 IP ozlabs.org.http > 192.168.5.101.50355: . ack 120 win 8192
This is good for programs like awk, but less optimal if you want to read it directly; then it's more interesting to be able to recognize the relative times, which vary by several orders of magnitude, from 69 µs to 2.2 s, a factor of 10,000 to 1. For that I prefer this format:
61.579152 IP 192.168.5.101.58187 > ozlabs.org.http: S 1157815307:1157815307(0) win 65535 <mss 1460,nop,wscale 1,nop,nop,timestamp 326554162 0,sackOK,eol>
1646 IP ozlabs.org.http > 192.168.5.101.58187: S 294010612:294010612(0) ack 1157815308 win 8192 <mss 1460>
69 IP 192.168.5.101.58187 > ozlabs.org.http: . ack 1 win 65535
119 IP 192.168.5.101.58187 > ozlabs.org.http: P 1:119(118) ack 1 win 65535
2549 IP ozlabs.org.http > 192.168.5.101.58187: . ack 119 win 8192
1.834737 IP ozlabs.org.http > 192.168.5.101.58187: P 1:360(359) ack 119 win 8192
9803 IP ozlabs.org.http > 192.168.5.101.58187: F 360:360(0) ack 119 win 8192
2142 IP 192.168.5.101.58187 > ozlabs.org.http: . ack 361 win 65535
19 IP 192.168.5.101.58187 > ozlabs.org.http: F 119:119(0) ack 361 win 65535
2225 IP ozlabs.org.http > 192.168.5.101.58187: . ack 120 win 8192
Ended up implementing both; here's the patch to tcpdump revision 3.9.
| Tuesday, 15 January 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Spent some time today writing up yesterday's patches and sending them off the specified mailing address, tcpdump-workers@tcpdump.org, only to get the reply that the address didn't exist: it's really tcpdump-workers@lists.tcpdump.org.
I finally need to get my finances and investments sorted out; that's not in the slightest helped by the positively useless web software available. I can't even easily get a display of the performance of my investments compared to what I paid for them, and at least one of the web sites is just completely useless at resolutions higher than 1024x768. Once again I'm completely amazed by the poor quality of web software; it looks like I'll have to write my own to keep track of what's going on.
| Wednesday, 16 January 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
As planned, spent most of the day going through budget stuff, with the interim result that we're roughly sticking to our overall budget, though the individual items are quite different from what we planned. In particular, we are spending nearly double what we planned on food and drink. More fine tuning needed.
The last time the AUUG web server was updated was 5 years ago. We had meant to do it at the hackers barbecue, but David Newall wasn't able to make it. Decided we had waited long enough for 6.3-RELEASE, so installed the latest ISO and upgraded to today's version of 6.3. Somehow I always seem to end up with a PRERELEASE tag.
| Thursday, 17 January 2008 | Dereel | Images for 17 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Into Ballarat today to finish the registration details for the other car and the horse float, which went well enough, but with that and shopping, somehow the day was over. Isn't it good not to have anything important to do?
Over to Chris' later, where they were having Mikey shod:
|
|||||||||||||||||||||||
| Friday, 18 January 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More working on finances today, reinforced by the arrival of the electricity bill. After installing the air conditioning and no longer using the electric heaters, and also after the arrival of more clement weather, we had expected a drop in the bill; instead it was 50% higher, an average power consumption of 2.5 kW. Did a bit of searching around, and it's fairly clear it has to be the computers. Each of them doesn't use much, but I have 9 of them running constantly. I need to so some thinking about how to hibernate them without stopping the nightly cron jobs.
| Saturday, 19 January 2008 | Dereel | Images for 19 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
We have rain! It's the first this year. It started yesterday evening, and by 9 am this morning 20 mm had fallen. Now why can't it be more regular?
Spent the day inside—the weather was almost like a European November—and addressed my photo pages to get back into practice with PHP. This was a particularly difficult effort: 15 years ago, in Porting UNIX Software, I wrote:
UNIX timekeeping is an untidy area, made more confusing by national and international laws and customs. Broadly, there are two kinds of functions: one group is concerned with getting and setting system times, and the other group is concerned with converting time representations between a bewildering number of formats.
Things haven't got better since then; with PHP I have another layer of date and time functions, none of which do what I want. The most irritating one is that the function strtotime() (which also, confusingly, handles dates) interprets a sequence like 4/3/2008 in US out-of-order sequence to represent not 4 March 2008, but 3 April 2008—and there's no way to override it. Spent some time adapting my code from the Black Box project to do the job, and now hopefully I can enter dates in almost any format into the photos page—clearly the US out-of-order format can't work, because it's ambiguous. But it's painful thinking out the Right Way to do things, and I'm sure there's something that I have missed.
A number of people got back to me on power issues. Edwin Groothuis suggested virtual machines, but that's a non-starter. The issue is that my machines all do something:
The issues are:
At first sight, eureka and dereel seem redundant, but I can't put 5 display cards in a single machine.
boskoop is a PPC machine, so that can't be emulated.
I could possibly emulate ugliness, but I'd much rather get rid of it altogether.
pain (also called eucla when running FreeBSD) is hibernated most of the time anyway, and I can't even use it to connect to the wireless network because it has some strange issue with Microsoft which makes it access the disk once a second. Nobody from the Microsoft space can tell me why or how to stop it, and I've already worn out one disk, so I'd rather leave it hibernated.
I need both ceeveear and teevee because of performance issues. Maybe a faster CPU would help.
All other machines are in individual locations, so I can't coalesce them.
One interesting consideration, which unfortunately won't work for me: Kenneth Tom contacted me and told me that he had moved his 9 machines to a single colocated machine, and saved a lot of money for power and network connectivity in the process.
| Sunday, 20 January 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
It's no longer as wet, but it's still pretty cool, so we spent most of the day indoors. Somehow spent the whole day updating web pages, answering a little mail, brewing and cooking, not doing any of it well. Did a roast beef with Yorkshire pudding, only to discover that I had misremembered the units of measurement; for future reference, a gill is a quarter of a pint (142 ml), not half. To my surprise, the pudding did set, but it wasn't edible. Also thawed out a “sauce” that proved to be full of vinegar. We need to mark these things more carefully.
AUUG still hasn't got round to doing a ballot of whether to continue or to fold. They're not even considering the option of dropping the membership fees. In the end, sent out another message to the auug-announce list. I wonder if anything will come of it.
| Monday, 21 January 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
A couple of weeks ago we were sweltering in the heat, but today was quite cool—despite weather forecasts, it barely hit 17°, so once again we spent the day inside. That's all the funnier because I heard from Arnd Bergmann that the temperature in Tübingen hit 15° yesterday. Strange weather indeed.
I continue to have problems with ECONNRESET status returned from remote network connections. I strongly suspect the BST implementation. Maybe there's a reason why it must disconnect if it gets into trouble, but you'd think it would be able to recover from that kind of problem if the encapsulated TCP could do so. In any case, decided to try running squid in a peer-to-peer mode, and discovered IRCache in the process, so connected up to that. I need to monitor what goes on there in more detail.
Watching TV in the afternoon. SBS has a series on cooking, Food Safari, that's relatively watchable, even if it contains some strange claims, like that Speck is a French word (and hearing the French chef de cuisine of an expensive Sydney restaurant say so, and pronounce it “spake” into the bargain):
Speck - made from the hind pork leg in a similar way to prosciutto, speck is boned before the curing and smoking processes begin. Its smoky flavour is a great addition to slow cooked dishes like boeuf bourguignon.
It's German, of course, and without any qualification it simply means “fat” (French lard), or maybe fatty bacon (which is normally called Frühstücksspeck). It's always fatty. What they describe above could either be Räucherschinken (smoked ham) or, if it has more fat than they describe, geräucherter Schinkenspeck, smoked fatty ham, and the description better fits the word jambon, which they also don't mention. It's difficult to be sure: the recipe for “Bœuf bourgignon” states:
300g speck, cubed
Any German reading that would assume what the French call lard de poitrine, raw, fat pork belly, or even belly fat with no meat at all in it, and indeed that's what Bocuse specifies for his recipe for bœuf à la bourguignonne in La cuisine du marché. It's also quite close to what I recall seeing in the programme, but it's not what's written on the web site.
Other unexpected ingredients are olive oil (“extra virgin”, of course)—it should be fat or butter—and the beef (“preferably Wagyu”, which Google tells me is a Japanese beef, a term limited to meat produced in Japan—nothing to do with the pointe du culotte specified by Bocuse or Madame Saint-Ange). Clearly the recipes need to be interpreted with caution. I don't know of any recipe for bœuf à la bourguignonne which includes anywhere near this proportion of smoked meat.
But then, maybe this nomenclature is just a further example of the rank stupidity of Australian producers who insist on inventing names or changing the meaning of existing ones. Here are some of the ones that particularly annoy me:
| Conventional name | Obfuscatory name | Comments |
| Chinese cabbage | Wombok | Seems to be a corruption of Cantonese wong bok |
| Spring onion | Shallot | Probably a correct term, but not in general use, and confusing |
| Shallot | Eschalotte | Another fantasy name, not quite French échalotte and also not quite German Schalotte, made necessary by the previous renaming |
Other recipes are also surprising: I've commented in the past that a biriani is one of the most difficult recipes I know. The SBS biryani recipe isn't; basically you just throw everything in, one component after another, which just seems all wrong to me. All the birianis that I know first boil the meat to create a broth (sometimes referred to as yakhni) in which to cook the rice, and before serving they steam (dum) the dish. But then, biriani is a whole group of dishes, and this one purports to come from Pakistan. I can't believe it would taste as good, though.
The real problem, though, is that the web site is, as usual, messed up. The overlap is not as bad as it was, but printing out the recipes is a disaster, resulting in 6 pt text for which you need a magnifying glass to read. Ended up downloading the ones that interested us and fixing them so that just the recipes remain, which was an interesting exercise in reformatting. Getting it roughly OK was very fast—just extract the middle section—but tidying it up to be standards compliant (most of these pages generate about 160 errors on the WWW validator) took much longer.
| Tuesday, 22 January 2008 | Dereel | Images for 22 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Didn't do much today apart from trying to recover Sunday's roast beef, in the process going into more details on the sauce brune of Madame Saint-Ange. That worked very well, but I managed to mess up the Yorkshire pudding again! This time I forgot the egg, and ended up with a kind of bread, which didn't taste as bad as it should have. The real breakthrough was warming up the beef. Conventional wisdom has it that you can't do it right, but with modern microwave ovens with “inverter” technology, very low heating levels work fine, and the result was hardly any different from fresh.
A number of people commented on my complaints about the rendering of the SBS web site. To be fair, it seems to have improved. It's still not right—the texts overlap in the titles, for example—but the most annoying thing is probably all the advertising on it, some of which also overlaps the text. Here a comparison:
|
||||||||||||||||||||||
|
||||||||||||||||||||||
If I had seen this improvement before I started, I probably wouldn't have bothered.
| Wednesday, 23 January 2008 | Dereel | Images for 23 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
I suppose it's a sign of the times that PhD students are now researching what my father was doing 50 years ago. A while back I got a message from Zuraini Md Ali, an architecture student in Sheffield, who wanted more information about my father's activities at the Kuala Lumpur Technical College (now Maktab Teknik) in the late 1950s, along with Julius Posener and Karl Voltz. Called up my father to talk to him about that, but ran into not completely unexpected problems: a couple of days ago, encouraged by both the improved reliability of my satellite connection and the electricity bill I had received, I powered off ugliness.lemis.com, which I had been using for the wireless Internet gateway, and redirected homephone.lemis.com, my Sipura SP 3000 ATA, to go over satellite. Technically that worked, but today was the first time I actually tried it out. The connection was almost unintelligible.
What should I do? The obvious thing was to turn on ugliness again, but that was complicated by the fact that it didn't have a monitor, so decided to move my laptop (pain or eucla, depending on the operating system) to its place. That required installing the software, of course, but that wasn't a problem: I had done it dozens of times before. This time, though, it didn't work; apart from these stupid messages “the wizard has detected another network connection. The modem may not work correctly”, it claimed to be connected, but ipconfig showed nothing, and there was no traffic. Decided that it might be due to the fact that the CD from which I installed was an older version, and there had been a firmware upgrade since then. Gave up in disgust.
Spent the afternoon writing up some recipes that I've been meaning to do for some time: Gazpacho Andaluz, Paella Valenciana and Cha Shao roast pork. After that I was feeling strong enough for another attempt at the VoIP stuff, recabled a monitor and fired up ugliness. It worked, of course. But what then? I didn't want to leave the machine running all the time, not just because of the power consumption, but because of the traffic it generates: for some reason it continually sends out NetBIOS name server requests for Chris Yeardley's server machine, even though I've disabled everything obvious in the configuration:
10:52:02.895681 IP CPE-124-186-14-82.qld.bigpond.net.au.netbios-ns > CPE-124-186-14-83.qld.bigpond.net.au.netbios-ns: NBT UDP PACKET(137): QUERY; REQUEST; BROADCAST 10:52:03.645755 IP CPE-124-186-14-82.qld.bigpond.net.au.netbios-ns > CPE-124-186-14-83.qld.bigpond.net.au.netbios-ns: NBT UDP PACKET(137): QUERY; REQUEST; BROADCAST 10:52:04.396905 IP CPE-124-186-14-82.qld.bigpond.net.au.netbios-ns > CPE-124-186-14-83.qld.bigpond.net.au.netbios-ns: NBT UDP PACKET(137): QUERY; REQUEST; BROADCAST
It's not quite once a second, but pretty close. Why the address I got is in the domain qld.bigpond.net.au. (which indicates that I'm in Queensland, thousands of kilometres away) is unclear, but in keeping with my experience with Telstra. Clearly I don't want that traffic, anyway. So how about suspending it?
The best thing you can say about the documentation is that it's better than Apple's. It explains relatively well the options (suspend is suspend to RAM, hibernate is suspend to disk)—but it doesn't tell you how to resume. Was able to suspend relatively easily, but it didn't power down the machine. Hibernation worked too, but the only way I could find to resume from hibernation was to press the power button, and that caused the system to spontaneously reset. What did I do wrong? Is this not the right way to resume from hibernation?
Once again asked on IRC, where people showed a complete lack of understanding of what I was trying to do. One went off in a long explanation of why I must reconfigure my window manager to pass Ctrl-Alt-DEL to ugliness, so that I could then push my mouse around on a selection window to shut the machine down, without first establishing that the window manager was the culprit, and ignoring my point that it was much easier just to press the power button. After a bit of messing around, established:
Somehow frequent retries are modern. After powering down ugliness, saw the following exchange:
18:13:30.636425 arp who-has ugliness.lemis.com tell homephone.lemis.com 18:13:30.636552 arp who-has ugliness.lemis.com tell homephone.lemis.com 18:13:31.636440 arp who-has ugliness.lemis.com tell homephone.lemis.com
And yes, this time they're every second.
It seems that there really is a word estivation, but it doesn't mean anything like the way I use it. I'll stick with Humpty Dumpty:
“When I use a word,” Humpty Dumpty said in a rather a scornful tone, “it means just what I choose it to mean—neither more nor less.”
My web server logs are interesting: my SCO code comparison page has been getting lots of hits. That usually means a reference in some web site recently, and since the collapse of SCO is topical at the moment, that's not very surprising. What is puzzling, though, is that nearly all the hits have no referrer. I wonder where it's coming from.
| Thursday, 24 January 2008 | Dereel | Images for 24 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Where is this traffic on ugliness coming from? Fired the machine up today to make a phone call, and saw the traffic increasing all the time. Fired up wireshark, which in itself is a pain: Microsoft doesn't seem to have the notion of interface names, so the interface selection window looks like this:
|
|
||||||||
It almost takes guesswork to work out which interface is which. What a pain!
The “connection manager” window was showing gradually increasing incoming traffic, not the outgoing NetBIOS name service traffic that I had already noticed; possibly this was being stopped by a firewall, since there was no kind of response. But then, since it's broadcast, that's probably normal anyway. So, what was the incoming traffic? There was none! This “connection manager” is lying. It's already clear that it's not in agreement with the statistics published on Telstra's web site, but they're not reliable either. And they're charging money based on this broken software!
More playing around with PHP today; I've started migrating my diary, so now it's diary.php instead of diary.html; the latter URL just redirects to the former. The first advantage of this change is going to be in the photos. Previously (and currently still in the monthly diaries) if you click on a photo, you get taken to the daily photo page. Now it shows the diary with a bigger image in line. What I do after that is something that I'm still thinking about.
Yvonne back from shopping today with a new mouse from ALDI. This was advertised as a mouse with 6 programmable buttons and 2 scroll wheels; that's incorrect: it has only 4 buttons. It's interesting because it's a wireless mouse without batteries; instead, it gets its power from an inductive pad on which it rests. An interesting idea. Under FreeBSD it worked out of the box, with the exception of the horizontal scroll button, which didn't seem to generate any output. I still need to get used to the idea, and every new mouse feels funny at first, but currently it doesn't look too bad.
Over to dinner at the Yeardley's in the evening. They're thinking of migrating their web site to PHP as well.
| Friday, 25 January 2008 | Dereel | Images for 25 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Over to the Yeardley's this morning to play around with PHP on the Narrawin site, with mixed success. delicious, Chris' Apple, runs Apache, but not PHP, so we had to load it to the live web site, where we at least confirmed that things worked. Now to decide how to use the new-found freedom, but also how to install the latest versions of Apache and PHP on delicious.
Back home, and still more playing around with PHP. I can see this keeping me occupied for months.
| Saturday, 26 January 2008 | Dereel | Images for 26 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Another week gone by, and it's hot again. The Bureau of Meteorology forecast 30°, but we got 34°. They also promised rain, and indeed we got a couple of drops, not even enough to wet the bottom of the rain gauge, and certainly not enough to stop the dam from drying out:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The first photo was taken on 10 November 2007, and the second today.
Today was Australia Day, the 220th anniversary of the arrival of the First Fleet at Sydney Cove, and there was a market at the Dereel Hall, just like there was on election day. It seems that that's a coincidence: the market's on the 4th Saturday of each month, and both days coincided with that. Nothing much to see at the market, so we went on to a nursery in Corindhap, where they're in summer survival mode. We should get back to gardening in a couple of months.
More work on PHP stuff. This is a can of worms. Also updated my bad language page with Microsoft's redefinition of the word “Internet”.
| Sunday, 27 January 2008 | Dereel | Images for 27 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
On 10 January 2008 I wrote a letter to Telstra complaining about their treatment and demanding satisfaction by 25 January. I received no response, so spent most of today documenting the problem, writing another demand—over $900!—and preparing a complaint for the TIO. What a pain!
| Monday, 28 January 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
On with the PHP work today, sort of. What I did had little to do with PHP.
One problem I currently have is that I'm not including the dimensions of images in the output HTML. It's not that I don't have them—they're there in the day's photos pages—but extracting them in the functions that I've been writing would be slow and tedious. Clearly what I need is a database, so today set to converting the photo information into a single MySQL table.
That in itself was fun: the information is in text form, like:
garden-se.jpeg 300 225 44 600 450 100 3648 2736 1792 house-nw.jpeg 300 225 46 600 450 102 3648 2736 1616
The values are triplets with the dimensions in pixels and the size in kB of the “tiny”, “small” and “big” images. Clearly a candidate for automatic conversion to a table, but there's one additional thing that I had to add, the name of the directory. And round about there awk began to run out of steam. Maybe I should really learn perl, but I've resisted for 15 years now, and I don't know if I can face it.
Probably the real issue is that there is almost no documentation for awk (really gawk). The man page tells you (badly) how to run it, but doesn't mention the language. The info page gives more information, but is even more badly structured than most info pages, and consists more of examples than of a language description. The book I use is sed and awk, by Dale Dougherty, published in 1990 and now presumably quite out of date. None of them told me how to find out the current working directory, or whether functions like basename() or dirname() are available. Ended up piping the awk output through a sed script and then into mysql, which worked relatively well.
In the afternoon, off by car to Enfield State Park (which Google Maps calls “National Park”) to look for some good riding terrain. It's a good thing we did: large parts of it are barren and have only particularly rocky tracks. In particular, the Berringa-Misery Creek Road and that part of Misery Creek Road between Misery Creek and Enfield live up to their names. In many ways it's less interesting than what we have directly at our front door. Finally found something starting round Dredge Road in Enfield.
| Tuesday, 29 January 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Yesterday the Linux.conf.au conference started just down the road in Melbourne. I had thought of going, but by the time I got round to thinking about it, tickets had been sold out, thus relieving me of having to decide whether to go or not.
In the mail this morning received an offer of a free ticket. Spent some time thinking about that, to the point where Yvonne packed a bag for me, but in the end I decided not to go.
Why? It's not that I wouldn't enjoy the conference. There are a couple of reasons:
In short, I'm getting old.
Continued my work on converting the photo pages to use a MySQL database. That worked surprisingly well; it's surprising how much easier it is to use MySQL than PHP, where I'm continually tripping over quote marks and things.
The real issue, though, is: how do I get this stuff on the external web site? MySQL replication isn't difficult, though I've never tried it quite like this, but I don't have access to the database on the external site. And then there are questions of naming. But this is the week of Linux.conf.au, and nobody's home, so it'll have to wait until next week.
| Wednesday, 30 January 2008 | Dereel | Images for 30 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More mail on my father's career today, including a request for his curriculum vitæ. Astoundingly, I found an old CV and was able to scan it in.
Also spent some time translating a recipe for rendang, one that uses no chili. It was anything but successful, maybe because the quantities were wrong, or maybe because there was something wrong with one of the ingredients: it ended up very bitter. More work needed.
In the afternoon, Keith Matthews posted a URL for a quiz on IRC. To my surprise, I got 100%:
Rasmus only got 96%.
And of course, then there was another one. This time things looked different:
That meant that Rasmus read this diary, and of course he immediately picked on my comments on PHP. First, my issues with PHP: I have stuff like this in photos/Photos.php:
<a id="' . $basephoto . '" name="' . $basephoto .
'"
href="Bigphoto.php?dirdate=' . $dirdate . '&image=' . $photo . '&width=100%">
<img alt="Image"
title="Click on the picture to see a larger version'
. " ($nextkb kB, $nextwidth x $nextheight pixels)"
. '"
<?php showphoto ("/grog/Photos/' . $dirdate . "/"); >>
I'm continually tripping over the ", ' and . characters. Rasmus suggested a couple of alternatives:
<Rasmus> groggy: Why are you tripping over quotes?
<Rasmus> they are pretty small
<groggy> Rasmus: Heh. I see you've been reading my diary.
<groggy> Rasmus: Because I'm not used to them, mainly.
<groggy> Rasmus: The idea of writing
<groggy> print '<a href="' . $url . '">';
<groggy> is still a bit of a fingertwister for me.
<groggy> Rasmus: What editor do you use?
<Rasmus> vim
<Rasmus> many ways to do that one
<groggy> Rasmus: Ah, pity.
<Rasmus> echo <<<EOB
<Rasmus> <a href="$url">
<Rasmus> EOB;
<Rasmus> would be one where you don't need to worry about it
<groggy> Rasmus: Ah, that sounds good. thanks.
<Rasmus> echo "<a href=\"$url\">";
<Rasmus> would be another
<groggy> Rasmus: Yes, that one I know.
<Rasmus> ?>
<Rasmus> <a href="<?php echo $url?>">
<Rasmus> as in, escape out of PHP mode..
<groggy> Rasmus: That one really freaks html tidy out.
<Rasmus> You should run htmltidy on the output, not on the template
<groggy> Is there a way of getting tidy to ignore PHP code?
<groggy> Rasmus: What good is that?
<groggy> Rasmus: I want to generate clean-looking HTML.
<Rasmus> because that's what the browser sees
<Rasmus> the browser doesn't see any of the PHP
<Rasmus> and anybody doing a view-source will see the html post-php
<groggy> Rasmus: Understood, and I do that too sometimes to convert to RSS.
<groggy> Right, but I see it beforehand.
<groggy> And that's what I use to generate the HTML.
<groggy> Besides, it's what I edit, and I'm probably the only person who
really looks at the source.
<Rasmus> tidy creates your html?
<Rasmus> how does that work? I thought it was just for checking it
<groggy> Rasmus: No, it can rewrite it too.
<groggy> Take a look at the HTML generated by diary.php.
<groggy> It should look relatively clean.
<groggy> I write stuff, then I have an Emacs macro that reformats it with HTML
tidy.
<groggy> But tidy gets confused by PHP (probably <?foo ?>) and indents
incorrectly.
<Rasmus> Ah, Emacs.. It has terrible support for multi-context syntaxes.
<Rasmus> and thus tidy can't distinguish either
<groggy> Nothing to do with Emacs really.
<groggy> Emacs just invokes tidy and then tidies up the result.
<groggy> But yes, the indentation with PHP and HTML in the same document is a
mess.
<Rasmus> throw in Javascript and CSS as well
<Rasmus> vim handles that perfectly
<Rasmus> 4 different modes in the same file
<groggy> Really?
* groggy looks at vim with renewed interest.
<Rasmus> yup
<groggy> Pity it would mean rewiring my fingers.
<Rasmus> it's syntax highlighting code is great
<Rasmus> -'
<groggy> So is there a simple command to vim to say "reformat this template
cleanly"?
<Rasmus> as in a built-in tidy? I don't think so
<Rasmus> It's more about multi-mode syntax highlighting
<Rasmus> although there is a tidy.vim rule thing
<Rasmus> I have never tried it though
David Yeardley over in the evening for dinner with a couple of his wives:
|
|||||||||||||||||||||||
Spent some time having fun and repeating some of the old tricks we did with Lilac:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
| Thursday, 31 January 2008 | Dereel | Images for 31 January 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Yet another day spent in the office. I'm supposed to be retired. I haven't even had a chance to watch TV for the past few days, and gradually my disks are overflowing.
Spent some time looking at my new ALDI mouse in more detail. Everything works except the horizontal scroll wheel, which xev doesn't notice. On Peter Jeremy's suggestion, tried running moused with the -d option. That shows quite a bit of information. When moving the vertical scroll wheel up, for example, I get:
moused: received char 0x87 moused: received char 0x0 moused: received char 0x0 moused: received char 0x0 moused: received char 0x0 moused: received char 0xff moused: received char 0x0 moused: received char 0x7f moused: assembled full packet (len 8) 87,0,0,0,0,ff,0,7f moused: tv: 1201759228 837111 moused: flags:80000000 buttons:00000000 obuttons:00000000 moused: activity : buttons 0x00000000 dx 0 dy 0 dz -1
The first 8 lines are of course the same as the contents of the full packet, and the final line shows what output moused generates. When moving the wheel down, the third-last byte changes from 0xff to 1.
moused sees the horizontal wheel too, but here we get (excluding the redundant first 8 lines):
moused: assembled full packet (len 8) 87,0,0,0,0,0,0,7f
moused can't do anything with this, presumably because the third-last byte is 0, no matter which direction I choose; in fact, the entire packet is the same as the vertical scroll wheel output except for the third-last byte. So there must be some initialization incantation for the mouse that nobody's telling about.
To be sure, tried it out on pain, the Microsoft incantation of my laptop. Interestingly, the mouse didn't work at all until I installed the driver. Then I fired up firefox and got the message:
|
|
||||||||
“The proxy server is refusing connections”. That's happened often enough, but in this case it was a downright lie. The proxy server didn't do anything, because the network cable wasn't plugged in. Sometimes I could scream. Why do people have to write such stupid, downright wrong messages? Whom does it help?
Plugged in the network cable, and of course everything worked, including the horizontal scroll wheel. But how do I find what's happening to initialize the mouse? A Google search revealed nothing useful. It would be really nice to have some kind of tcpdump tool for USB.
Do you have a comment about something I have written? This is a diary, not a “blog”, and there is deliberately no provision for directly adding comments. It's also not a vehicle for third-party content. But I welcome feedback and try to reply to all messages I receive. See the diary overview for more details. If you do send me a message relating to something I have written, please indicate whether you'd prefer me not to mention your name. Otherwise I'll assume that it's OK to do so.
| Top of page | Previous month | Greg's home page | Today's diary entry | Next month | Greg's photos | Copyright information |
