|
| Greg's diary |
| May 2008 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Thursday, 1 May 2008 | Dereel | Images for 1 May 2008 |
| Top of page | ||
| next day | ||
| last day |
Finally my new lens is there! I've been agonizing over buying a telephoto lens for over 9 months, and the near-impossibility of buying lenses on eBay from US added to the problems. Last week I finally bought an Olympus ZUIKO DIGITAL ED 70-300 mm f/4.0-f/5.6 from a seller calling himself ace of bicycle, along with a second-hand camcorder from a different vendor. The lens was located in Tokyo, the camcorder in Melbourne, so when I got a note from the post office on Monday that a package was waiting for me, I assumed it was the one from Melbourne. Not so: I'm still waiting for the camcorder, but the lens arrived on Monday, and Chris picked it up from Sebastopol Post Office yesterday afternoon:
|
|
||||||||
The lens is interesting because it can focus as close as 0.96 m; this makes it interesting for taking macro photos. Here my mystery tree with the yellow flowers; the individual flowers are about 5 mm across:
Into town for multiple purposes today: bought a new electric drill, the first in over 20 years, and also bought some new shoes. For as long as Yvonne has known me, I have been wearing sandals made by Mike Marquez in Los Gatos, but he moved on years ago, and my final pair has given up, so I'm reduced to wearing normal sandals. Also bought a single rose bush (“Lili Marleen”), which we planted outside the entry to the kitchen.
| Friday, 2 May 2008 | Dereel | Images for 2 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Quiet day. Autumn is well upon us, and we didn't feel like doing much outside. Spent some time upgrading my PHP scripts for making web pages out of lists; it's amazing how many things can be handled that way, for example my garden index is made from a relatively simple list of URLs.
One thing of interest was the mystery flower I mentioned yesterday. Callum Gibson identified it as a Lantana, and indeed it looks likely. The following links show photos all quite like ours: ausgarden.com.au, labouichere.com, Wikimedia commons, sunvalleylandscape.com and integritylandscaping.com. This one, from Wikimedia, is closest to ours:
The trouble is that Lantana is a weed of (Australian) national significance. The Queensland Department of Primary Industries has a fact sheet. On this page they also state:
Stems are square with small, recurved prickles.
Leaves are bright green, about 6 cm long, with roundtoothed edges and grow opposite one another along the stem.
Flowers vary in colour from pale cream to yellow, white, pink, orange, red, lilac and purple, about 2.5 cm in diameter.
Fruits are glossy, rounded, fleshy, purplish-black when ripe.
I've tidied up the punctuation somewhat, but there are a couple of differences here: the stems of my plant have no prickles, and the plant has no fruit. But prickles in themselves don't seem to be a very good indication of the genus, and elsewhere I read of “sterile” Lantanas. That might sound like we're out of the woods, but then I found an ASGAP article which states:
The Newsletter of the Environmental Weeds Management Group (EWMG) (Oct. 2001) notes evidence of even so-called 'sterile' garden varieties of lantana producing pollen which may cross-pollinate wild lantana and produce new varieties in the wild.
Did a bit of thinking about that. We don't seem to have any wild Lantana in the area, and even if there were, wouldn't the chance of cross-pollination tend to produce more sterile varieties? Thought about that for a while, without coming to much of a conclusion. Then read Allan Seale's “Australian Gardening”, a book which dates back to 1985, and which states:
Lantana. Long-flowering and drought and heat-resisting shrubs for all but the coldest regions—easily managed and should not be confused with the noxious wild L. Camara.
So what to do? While I was thinking of this, Callum came back with another suggestion: now it's a Buddleia (or is that Buddleja?), and indeed that looks likely. Here's a photo from Banwy Valley Nursery, which describes it as a Buddleia ‘Sungold’, followed by one of the photos I took yesterday:
|
|||||||||||||||||||||||
The domain of Banwy Valley Nursery is no longer registered to them, and the image is gone. That's not important: in the meantime I have established that this plant is a Buddleja globosa. And no, it's not invasive.
Still later I discovered that this is in fact a Buddleja × weyeriana, presumably cultivar “Sungold”, so the original appears correct
We're not done yet; it seems that Buddleja also has potential to be invasive. More reading to do.
| Saturday, 3 May 2008 | Dereel | Images for 3 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More investigation of the Buddleia issues today. Yes, there's a Buddleia Davidii that is considered a noxious weed, but that appears to be quite different from the one I have, for which I still haven't found a name. Mine is not on sale in Australia; I wonder if that is an indication that all Buddleias are considered dangerous. I don't see any reason to believe that mine is. The June 2007 newsletter of the Monarch Butterfly NZ Trust states:
There's a lot of confusion about Buddleia. Ask for it at your local garden centre, and they'll probably tell you “no, it's a weed”. But it is only the B. davidii that is listed as a plant pest — and even there, only the mauve flowering version that causes the damage as it multiplies in bright profusion.
B. davidii has been declared a danger to our primary industry, as it breeds prolifically, creating problems in pine forests. But there are still several Buddleias that are permissible and don't create problems (they don't seed, they are sterile). They are great nectar plants.
In fact, the MBNZT has been entrusted to trial a new cultivar, B. Silver Anniversary, to be released later this year; we have been asked to measure how successful it is providing nectar for our butterflies — and bees too. “Silver Anniversary” has clusters of white flowers with mustard coloured eyes and a sweet honey scent.
Of course, the Monarch Butterfly NZ Trust has vested interests, but they can't be that far off the mark.
Didn't do much else during the day. I've established that the irrigation system in the garden does need two separate sections for the eastern part, as I had originally planned. Annoyingly, the pressure drop in the system means that I can't use up the supply capacity of the pump and still have adequate pressure, so it cycles anyway. I can't see a good solution for that one. Maybe I should turn both solenoids on at the same time.
Over the last couple of weeks, a group of people in the UNIX Heritage Society, led by Warren Toomey and Tim Newsham, have scanned in an old paper copy of the First Edition UNIX kernel and got it to run. The kernel is dated June 1972, and it's written in PDP-11 assembler. It's really strange to see code like this:
:login: root root # ls -l total 6 43 sdrwr- 2 bin 620 Jan 1 00:00:00 bin 42 sdrwr- 2 sys 250 Jan 1 00:00:00 dev 44 sdrwr- 2 bin 110 Jan 1 00:00:00 etc 46 sdrwrw 2 sys 30 Jan 1 00:00:00 tmp 45 sdrwr- 2 sys 20 Jan 1 00:00:00 usr # chdir bin # date Fri Jan 1 00:00:24
It's also amazing how quickly it has happened; Warren started things rolling on 23 April when he mentioned the document on the TUHS mailing list. There's now a new mailing list which deals with exactly this reconstruction. I found the most interesting part the issue with executable binaries: in the First Edition, the a.out header was 12 bytes long, and from the 2nd Edition it was 16 bytes long (which it still is today). This caused some problems because the earliest available version of many binaries was from the Second Edition. But the really interesting thing is the meaning of the magic number: in the First Edition version it's 0405, and in the Second Edition it's 0407. Each is in the first word of the object file, and it doubles as a PDP-11 instruction to branch over the header (5 or 7 words, meaning 10 or 14 bytes, added to the two of the instruction itself).
| Sunday, 4 May 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More work in the garden, in particular the slow and boring digging up of all the garden soil. I can't make up my mind whether to spray the weeds and let them die first, or just dig them up. One way or another, it looks as if we're going to be busy with the project for months.
Finally got round to replacing the power supply in the old teevee. That's just the start of what threatens to be a significant rearrangement of my hardware. In particular, my backup disk is overflowing, so I'll have to replace it with a larger one.
| Monday, 5 May 2008 | Dereel | Images for 5 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Finally the video camera that I bought on eBay two weeks has arrived. The seller had waited for PayPal to pay the money to him before sending it, which he did in a barely adequate padded envelope. The battery seems to be dead. Do I complain, or just capitulate?
I'm still looking for long telephoto lenses on eBay. In the process, managed to lose a lens that I was “watching”: it didn't show up in the listing to which it belonged. After some examination, discovered that there are two completely different eBay sites which purport to show the current auctions, but they only barely overlap in their content. The following two images show the same search done at the same time on http://search.ebay.com.au and http://cameras.listings.ebay.com.au:
|
||||||||||||||||||||||
|
||||||||||||||||||||||
I'm amazed. I'll keep an eye on it and see if the discrepancy continues.
Finally got around to putting the spare 250 GB disk into teevee, in replacement for the 200 GB disk that was there already—only it wasn't a 200 GB disk, it was 250 GB. Further investigation showed that the default newfs parameters for UFS 2 are less than ideal nowadays: the “free-space” is 8% ( 20 GB), and there's one 512 byte inode for every 4 * frag-size bytes (implying that this is the average size). frag-size is 2 kB, so that corresponds to over 6% of the total space. Looking at my file systems, it's clear that this is particularly wasteful:
=== root@teevee (/dev/ttyp4) ~ 8 -> df -i
Filesystem 1048576-blocks Used Avail Capacity iused ifree %iused Mounted on
/dev/ad0s1a 9916 6722 2400 74% 539201 779709 41% /
/dev/ad0s1d 8044 3022 4378 41% 3332 1080058 0% /home
/dev/ad1s1d 184772 139816 30174 82% 125 24470401 0% /spool
/dev/ad3s1d 238422 190980 45057 81% 616 67414 1% /spool2
eureka:/ 14873 12017 1665 88% 665333 1313033 34% /eureka
eureka:/home 9916 4766 4356 52% 171949 1146961 13% /eureka/home
eureka:/src 188356 139071 34216 80% 6699765 18221001 27% /src
eureka:/dump 122037 122037 0 100% 23 16156647 0% /dump
dereel:/ 13868 6531 6228 51% 550693 1286361 30% /dereel
dereel:/home 116092 82025 24779 77% 308545 15070909 2% /dereel/home
dereel:/dumpa 76285 71518 0 100% 330 23220 1% /dumpa
dereel:/dumpb 187780 182270 3632 98% 485 12218393 0% /dumpb
In particular, the dump file systems and the /spool file systems on teevee both have an average file size of 450 MB, more than 5000 times the default. Ran newfs on the new disk, with the parameters -U -m 1 -i 50000, which enables soft updates, leaves 1% free space, and creates one inode for every 50 MB (I thought). In fact, looking at the output above (for /spool2), it seems to have created one inode for every 3.5 MB. Is this a bug? To be investigated next time I run newfs; I certainly don't want to have to move 190 GB just to try it out again.
More work in the garden. I suppose I should only mention that when I do something I want to recall.
| Tuesday, 6 May 2008 | Dereel –> Melbourne –> Dereel | Images for 6 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Chris Yeardley left for the Gold Coast this morning, and we had been intending to go to Melbourne for some time, so took her to the airport, then on to Palm Place Nursery, not far from the airport, to buy a Bottle Tree:
|
|||||||||||||||||||||||
The one we bought wasn't the one in the middle, nor even one of the little ones in front of it, but an even smaller one on the other side of the display. In about 10 years it might be interesting. Also bought a Cycad for Yvonne and a Bird of Paradise flower for myself.
Then down Sydney Road to the same Italian supermarket we first visited last winter. On that occasion we didn't know what to expect at the Victoria Market, and we didn't look too carefully. This time we did. I suppose if you're Italian it's great—I heard the staff talking to the customers in Italian—but for somebody with just a casual interest in Italian food, it didn't seem to be quite what we were looking for.
On to the Victoria Market, where Yvonne bought lots of meat and sausages, and I found an “Asian” supermarket which had most of the stuff we would otherwise have bought at the Footscray Market. Off towards Grain and Grape, when we felt hungry, so down to Footscray anyway for a Vietnamese soup and noodles. Then to Grain and Grape, where I bought 60 kg of malt—that should keep me going for a while.
Back home through pouring rain—8 mm since we left, more than we've had in a single day in over a month. Somehow we were rather tired in the evening. Must be the age.
| Wednesday, 7 May 2008 | Dereel | Images for 7 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Another day with little to show. Baked some bread, planned a shed for the horses, and did some inevitable work in the garden, including planting two of yesterday's plants (the Bottle Tree and the Bird of Paradise flower). We're still trying to work out where to put the Cycad. Somehow, despite its size, our garden is filling up.
In the garden we found some tell-tale tracks:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The kangaroos are back, and they've been chewing on our plants. For some reason they like Heartsease (the remains of which are shown above). They also like acacias, and they've had several attempts at a couple that are still barely surviving. Decided to put some old plastic drink bottles to good purpose:
|
|||||||||||||||||||||||
Fresh sardines for dinner, which we did on the barbecue. That required a new cylinder of gas: our old one had run out a couple of days ago, and the spare I thought we had was nowhere to be seen. It's probably a sign of the culture that there's no petrol station between Napoleons and Rokewood, but the Dereel General Store (phone 5346 1391, a number we had difficulty finding) has barbecue gas.
Sardines taste good, but the preparation is rather intensive:
|
|||||||||||||||||||||||
| Thursday, 8 May 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Another day with little to report. My email inbox is getting larger all the time, and this morning I had nearly 5000 messages in it. I had intended to get rid of 2000, but in the end didn't quite make it. It leaves me wondering, though, how long it would take somebody with a monstrosity like Microsoft “Outlook” to process 2000 messages, even just to delete them.
Finally chopped down the fruit tree to the north-east of the house, the one that was too cramped. It's a pity; the fruit must have tasted wonderful, and it's sad to have to chop down something of that size, but it really didn't fit. Now we have to decide what to do with the area.
Telstra has finally paid up in full! Or at least they tell me they have: it'll be credited to my phone bill. I had to write three letters to them, none of which were answered (this one just said “Custumor feedback such as this, whilst [sic] unfavourable, is appreciated”). The fact that it was the writer's own incompetence is, of course, not mentioned.
| Friday, 9 May 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Getting the horses under cover for the winter is becoming a priority. Last winter (after moving to Dereel) we had no shelter, and Yvonne had to rug them all through the winter. We still don't have the money to build a real shed for them, but we have one of the old dog sheds standing empty. It's high enough, so we're planning to fold out one wall by 90° as a wind break, providing an entrance for the horses in the process, and remove the innards of the shed. The biggest issue is that the rafters are propped up by small posts across the width of the wall, so we're going to have to replace them by a beam. Into town with the Yeardley's big trailer to buy wood for that, and also used the opportunity to buy some shelving for the cupboards in the house, many of which, 10 months after we moved in, are still full of packing boxes.
Australia officially adopted the “metric system” in the early 1970s, more than 30 years ago. This means that all our boards are measured in metric units: 3.6 m, 4.5 m, 5.4 m. Always a factor of 0.9, the difference between a yard and a metre (4, 5 and 6 yards in this case). And I saw some electric saws on sale today with a 254 mm blade (10 inches). When will we really go “metric”?
| Saturday, 10 May 2008 | Dereel | Images for 10 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Not much garden work today, for a change. Instead spent some time trying to install shelving, progressing at a snail's pace. At this rate, I calculate it'll take me two months. But of course it won't: most of the time was just spent looking for tools and working out how to do the job. On the positive side, found a circular saw that my father must have brought from Bendigo years ago. It still works and is a great help.
Darah's Greasy Heel (or whatever it really is) is still not healing up. We had thought that it was sensitivity to light, so we bandaged up the legs; but it's now getting worse again, not better, despite treatment. Looks like yet another visit to the vet—it's cost us over $1000 already.
It's only been a few weeks since I last complained about the SBS web site. Now they've managed to has mess it up again. If this link doesn't show a PHP stack back trace, there's a copy here.
It seems that the hp3 referred to in the message is the end of the URL. Now you can get to it via the link without the file name part, but that doesn't seem to be the correct one. With a bit of searching I found my programme at http://www.sbs.com.au/schedule/2008-05-10/SBS%20Regional. They've also frobbed it so that a downloaded copy of the original no longer displays correctly. On the positive side, they have reinstated the hidden details—but only for the evening programme. Clearly somebody in their programming team thinks that the rest of the programme isn't important enough to describe by default. Sheesh.
In that connection, it's interesting to note that NineMSN (yes, “MSN” stands for “Microsoft Notwork”; enough said) sued IceTV for publishing details about their programme. IceTV won in the first instance, but they lost the appeal.
But why sue them in the first place? You'd think that a TV station would want people to watch their programmes, and they're clearly incapable of describing their programme themselves: they go one step further than SBS and require you to select a separate page for every programme. On my admittedly slow connection, it takes 10 seconds per programme, only to frequently discover there is no description. And of course the pages are not standards compliant and render incorrectly on my screen:
|
|
||||||||
Presumably they're intended for Microsoft users only.
The article above states:
Most industry observers speculate that Nine has vigorously fought this action to diminish the power of viewers effectively building their own programming, recording and skipping shows via EPGs -a significant threat to advertising revenue.
That's quite possible, of course, but in my opinion it shows a basic misunderstanding: Nine isn't the only channel broadcasting in Australia, and the programme isn't good enough that people would normally visit on chance—arguably, in fact, it's the worst of the lot. So by not advertising their programme, they're losing, not gaining.
| Sunday, 11 May 2008 | Dereel | Images for 11 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More work on the shelving today. I finally finished the framework for the first shelf! That sounds like nothing, and it is, of course. As usual, it wasn't the work, it was finding out how to do it with the tools at my disposal and coordinating it with other things I need to do.
One of these days I'm going to write a rant about plasterboard. On the right side of the cupboard there's only one stud, so the frame needs to be attached to the plasterboard. And this stuff is so soft! Spent quite some time before I found a combination of screws and plugs that would hold in the bloody stuff.
Finally got things finished, but then it occurred to me that this is the cupboard where I need to bring in cables from under the house, so I can't go any further. Now I need to pluck up my courage and crawl under the house. Mañana, at the earliest.
| Monday, 12 May 2008 | Dereel | Images for 12 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The final draft for “Beautiful Architecture” is due in on 15 May, US time (in other words, 16 May), and I spent most of the day working on that, mainly to consider the feedback that I received a month ago. The main question is: when describing a system architecture, do you just describe the things that make it special, or all things that differ from the norm? The Tandem/16 had an interesting instruction set and memory model. It didn't make it any more reliable, but I feel that it's an interesting thing to include in the chapter, while the editors thought it might be off-topic.
| Tuesday, 13 May 2008 | Dereel | Images for 13 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
CJ along today to help us convert the dog shed into a horse shed:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It kept us going all day, and we didn't quite get finished, but considering the quality of the material, some of which was completely rotten, I was quite happy with the result. The shed was obviously once a cowshed: CJ recognized some of the equipment that was still there.
| Wednesday, 14 May 2008 | Dereel | Images for 14 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Yesterday we did the first 80% of the shed, so today we did the other 80%. At least we got finished:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Doing it ourselves (with CJ's help, of course) saved us roughly 80% of the quote we had. It was pretty tiring, though—something I'll have to get used to.
The bloody kangaroos are here again! They came within a metre or two of the house, to judge by the droppings, and ate most of a grevillea we had planted in the north-west bed. We'll have to put in more protection.
| Thursday, 15 May 2008 | Dereel | Images for 15 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Today was the day I was supposed to finish my draft for “Beautiful Architecture”, but somehow I didn't even get started. Instead spent most of the day in the garden.
We've made some progress in identifying the plants in the garden. CJ's friend Sue yesterday identified the orange flowers as marigolds, and also the blue daisy-like flowers as everlasting daisies. On examination of the evidence, I think she's mistaken on the latter, though the stuff is tough as old boots. Last month I dragged out some of them while reorganizing a garden bed, and left them lying in front of the bed:
|
|||||||||||||||||||||||
That was nearly 3 weeks ago. Now there's a flower on the thing, though it has been out of the ground all that time:
|
|||||||||||||||||||||||
We didn't want to put the horses in the shed last night, because it was getting dark. Instead we put it off until today, where we put them in just as it was getting dark:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
| Friday, 16 May 2008 | Dereel | Images for 16 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Today was 15 May in the USA, my deadline for “Beautiful Architecture”. I made it, but the attempt to convert to DocBook didn't. Somehow I remain at odds with modern attitudes to interfacing to computers. O'Reilly states in their description of the virtues of DocBook:
... typesetting markup like LaTex is inherently focused on formatting—font size, margins, alignment, etc. We'd rather you spent your time focused on the semantic structure of your book (how sections are divided, which information goes in a sidebar versus a note, and so on), and that's where DocBook shines.
Well, maybe. There are things like code examples, file names, user input and many more, and they need to be mapped to a relatively small number of styles (file names in italic, user input in bold monospace, code examples also in monospace, and so on. If you want to change one of these representations, DocBook gives you the flexibility to do it easily.
The problems that I see are:
The overly pedantic syntax. In particular places, only particular tags are allowed. I haven't examined DocBook as much as XHTML, but in the latter case I'm continually running into problems where what I want to do is in conflict with a relatively abstract syntax definition. For example the photos on these pages, which I've mentioned in the past: you can't put a table in a paragraph. Why not? Because the syntax says so.
The inability to frob things manually. Widows and orphans are the bane of every typesetter's life, and the only method I know is to manually change the markup to (for example) do a premature page break. That's contrary to the spirit of DocBook, as many books based on DocBook can testify.
I have a different viewpoint: you shouldn't be able to recognize the typesetting software from the result. This is one of my biggest objections to LaTeX, though to be fair that's mainly due to the use of the default fonts.
I can't find a way to include images and drawings, something that I can do in groff with direct PostScript imports and tbl respectively. This reduces DocBook to a partial solution.
Many people say “but that's the cost of progress: you shouldn't have to read the document source. UNIX is old-fashioned. Use the XML editor”. But that's missing the point. When UNIX came on to the scene, non-user-readable documentation was already well established. UNIX showed that it was easier to understand what goes on under the hood.
XXE is O'Reilly's “editor” of choice. It apparently does a good job of validating the XML, something that HTML tidy was not able to do on my attempts at conversion (another sign of the complexity of the syntax?), but it requires you to use a mouse click to choose every tab—is that progress? It also has a character select window similar to the one I complained about in my paper “But is it UNIX?” five years ago (page 26 of the slides). My objections remain.
Yes, you don't have to use XXE like this. elsewhere the documentation tells you how to really use it. But you still need to relearn all the key bindings, and that's a serious drawback. Of course, there's also nXML mode for Emacs, so it's not really an issue, except that XXE is O'Reilly's tool of choice.
| Saturday, 17 May 2008 | Dereel | Images for 17 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
What miserable weather, and just what the doctor ordered for the clinic! In the 12 hours to 9 am this morning we got 18 mm of rain, and it continued pretty evenly all day, dropping another 10 mm by 17:00.
Spent most of the day learning about DocBook. Some of the things I claimed yesterday seem to be incorrect, at least in theory. Started reading Norm Walsh's book on DocBook, now for version 5.0. I suppose there's some significance in the fact that the first couple of chapters talk about structure and syntax. Chapter 3 starts with the interesting statement:
A key feature of XML markup is that you validate it.
What does validate mean? I validate my documents by formatting them and confirming that the output is what I want. But that's not what it means here. It means running the source through a validation program to confirm that the syntax is correct. I'm reminded of the quote attributed to Donald Knuth:
"Beware of bugs in the above code; I have only proved it correct, not tried it."
But why validate the syntax? I'm trying to write a book, not confirm that my syntax is correct. Compare this to human language: all languages have syntax, but how many people understand the syntax of the language they speak, let alone validate their syntax? Syntax should be something that comes naturally, and a good language tolerate will minor faults in the syntax, unless they get in way of understanding. That last sentence, for example, contains two syntactical errors, but I don't think anybody will have trouble understanding it.
Not so DocBook: the chapter above goes on to say:
The DocBook schema is a precise description of valid nesting: the order of elements, and their content. All DocBook documents must conform to this description or they are not DocBook documents (by definition).
It doesn't claim that they're easy to understand or use. Not surprisingly, most of the content is about the structure. After reading three chapters of the book, I still don't know what tags to use to mark up my document, nor how to format it. About the only thing I have picked up is that any component (file) of a document has its very specific place in the hierarchy, and it can't exist elsewhere. Does this mean that to test my markup I need to format the whole book? I don't know yet; I've only read three chapters.
This strict hierarchy reminds me of my first exposure to databases, in 1973, when I had to write a specific query language for a COSADYL hierarchical database. Relational databases came only a little later; they don't seem to have made it to document preparation yet.
Also spent some time looking at the nXML mode, about which I saw glowing recommendations on one of the O'Reilly links; now I can't find them any more. I also can't find any real documentation, though maybe that's partially a problem of the FreeBSD port; but there's none on the web site, either. As far as I can tell, its main purpose is to show what tags you can use where, done by pressing C-Return. But that doesn't work for me; instead I always get the message Cannot complete in this context. Maybe it's related to the comment I get when I open a file: Using vacuous schema, which suggests to me that the port hasn't installed some nameless component.
It might be worthwhile to follow up this further, but from the mode help I see that nearly all the commands relate to inserting tags, and somehow that's not enough. I might just as well use my HTML markup macros, which do things differently, but I'm used to them.
Discovered some anomalies in my PHP scripts for displaying photos. In the “small” versions I print excerpts from the EXIF data when it's of interest (for example, exposure compensation, but only when it's not 0). That seemed to work for 9 February, but not for 16 February. Further examination of the corresponding EXIF files (created by my Olympus E-510) showed, apart from the obvious differences of date and exposure:
=== grog@dereel (/dev/ttypi) ~/public_html/Photos 4 -> diff -wu 20080209/house-ne.exif 20080216/house-ne.exif
--- 20080209/house-ne.exif 2008-02-09 12:05:36.000000000 +1100
+++ 20080216/house-ne.exif 2008-02-17 11:05:57.000000000 +1100
-x-Resolution |72.00
+x-Resolution |314.00
-y-Resolution |72.00
+y-Resolution |314.00
-Exposure Bias |0.0
+Exposure Bias |0.00 EV
-MaxApertureValue |3.61
+MaxApertureValue |3.61 EV (f/3.5)
-InteroperabilityVers|
+InteroperabilityVers|0100
This is a diff extract: the lines starting with - refer to the first file (9 February), and the lines starting with a + refer to the second file (16 February). For the purists, I've also rearranged the lines to make the differences clearer. But why the difference? I've already noted bugs in the EXIF output of the Olympus E-510, but these ones don't seem to make much sense. I suspect it depends on whether I have used the AUTO setting or one of the others; I changed a couple of months ago. But there's nothing in the EXIF data to show that change. More investigation needed.
To the Yeardley's for the clinic dinner in the evening. Everybody was very tired; must have been the exertion of spending the day in the rain.
| Sunday, 18 May 2008 | Dereel | Images for 18 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Talking with Warren Toomey on IRC today about the First Edition of UNIX, and spent some time discussion the calling sequences. As I had commented in my chapter for “Beautiful Architecture”, the PDP-11 was not a real stack architecture. In particular, there's no frame pointer, and the subroutine call (JSR) was quite strange by modern standards: it saved the return address in a register, not on the stack. It saved the old contents of the register on the stack. The calling sequence used in the first edition of UNIX was something like (from ls.s, the first version of ls(1):
jsr r5,questf; < nonexistent\n\0>; .even
This placed the string " nonexistent\n" immediately after the call, leaving the return address in R5. It was up to the called function to determine the end of the string (then as now delimited by \0) and return to the address after it, rounding up if necessary to an even address.
Other variants included things like this, from ld1.s, part of ld(1):
jsr r5,oattach; otrelp; datsiz
Here there are two parameters, otrelp and datsiz. Clearly it was up to the function to know or calculate the length of the parameter list. Presumably a call like this would completely confuse things:
printf ("%s %s %s\n", foo, bah);
printf would happily print out three values, including the first instruction word of the code to which it should return, and then return to the second word.
Decided that it was time to finally do my duty to Callum Gibson and RFC a fix I had committed for him some time ago (months?). That meant that I needed a running version of FreeBSD 7-CURRENT, so set to installing that. Removed the /usr/src directory, since I needed the latest version, and watched it go forever. It wasn't until some time after I stopped it that I realized I had mistyped: instead of
=== root@swamp (/dev/ttyv0) /usr 195 -> rm -rf src
I had written:
=== root@swamp (/dev/ttyv0) /usr 195 -> rm -rf /src
By the time I realized what was happening, I had deleted about 40 GB of source files on my main source disk. Fortunately most of it was backed up, with the exception of the stuff I had been looking at with Warren this morning, so spent the rest of the day rebuilding the disk.
On the DocBook front, gave up with nXML and decided to install XXE in the hope that it would do better. The first steps were anything but encouraging:
|
|
||||||||
The error in question was by no means, as you might assume from the error window, that there was a missing <sect1> tag, but that there was a missing end tag, </sect1>. This is a fatal error; it refuses to have anything to do with the document. That must say something about the state of the art of DocBook: the syntax is so complicated that even the chosen tools can't fix it. Imagine a program editor that wouldn't edit programs with syntax errors in them!
Went back to Emacs, where I discovered that nXML does recognize the syntax errors, and puts a single red underline under two of the characters:
|
|
||||||||
If you position on this underline (how?), you get an explanatory message, in this case Missing end-tag "para". Now wouldn't it be nice for it to put it in?
Spent a couple of hours going through the document, one error at a time, until I finally got XXE to accept it. It still doesn't like it, but it saved an “invalid” document, reformatting it in the process—mercifully without reducing paragraphs to single lines, but not quite the way I would like it, and also not the way nXML likes it. Some choices are so strange that I think they're more like bugs than features. Why does XXE put a <sect2> header at the end of the previous paragraph?
|
|
||||||||
I don't understand what's wrong with the document, but the paragraph numbers are displayed out of order on a display that looks something like a web browser. It also shows comments and things as well, which makes it marginally useful. It's a pity that it doesn't have real editing functions.
And DocBook itself? I still can't make friends with it. The choice of the characters < and > for delimiters was very unfortunate, and makes for a lot of work converting normal documents, as well as making the source even less legible, (though still not as bad as TeX, which has similar problems with delimiters.
| Monday, 19 May 2008 | Dereel | Images for 19 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Another quiet day. This cold weather is making it less attractive to work outside, but that doesn't explain why I didn't do much “work” indoors either—the shelving is still waiting for me to continue. Instead, spent most of the day at the computer, at least committing a couple of MFCs and reviewing a patch. But then, that's the joy of being retired.
The patch was interesting: ping has an “audible” option, -a. If you set it, it creates a beep for every packet that is returned. There's also an -A option for the opposite: beep if a packet is not received before the next one goes out.
ping6, from the KAME project, doesn't have this facility. Instead, -a has been used for something completely different:
-a addrtype
Generate ICMPv6 Node Information Node Addresses query, rather
than echo-request.
How do you handle this kind of incompatibility? Shoot the developer? The patch incorporated both semantics, deciding on the meaning depending on whether a addrtype parameter was supplied. To my surprise, it worked, but it had to be ugly, and I don't see it making it into the FreeBSD project like that. My best suggestion is to rename the current ping6 -a option (why -a, anyway?) to -Q (-q is already taken).
| Tuesday, 20 May 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
CJ along again today to help put up some fencing, not helped by lack of materials—Yvonne set off twice with Sue to look for stuff at CJ's place, and once to Ballarat. Things got rather late as a result, and we stopped in almost total darkness after getting the fences into a condition where we could let the horses in; there's still plenty to do.
I'm giving another presentation of my brewing temperature control system at Grain and Grape on Saturday, so set to updating the slides. To my surprise there was a serious bug in the software, though the file was dated 12 May 2006, when I presented it in Ottawa. Spent some time fixing that; I still haven't got my laptop to display both on the internal and external displays, something lost when my laptop disk died in August 2006.
Playing around a little more with XXE, mainly trying to get it to display in something larger than flyspeck. Didn't manage that, but did find some other stuff of interest. It has the option of displaying a syntax diagram on the left. This is just a small part of the screen, but the flyspeck ensures that the thumbnail version is illegible:
|
|
||||||||
Clicking on it makes it better. Clearly there's a lot of new stuff to get used to, but it has the typical “one window” approach with lots of scroll bars. Why do people still do this? Multiple windows, individually resizeable, have been available for decades. And I still can't get used to this almost religious attention to syntax.
| Wednesday, 21 May 2008 | Dereel | Images for 21 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
In the summer we had plenty of grass in the paddocks, and to get rid of it we got Chris Yeardley to bring in some of her horses. That proved to be a mistake: the autumn rains didn't come, and now we have almost nothing in the paddocks. For the first time I can recall, the horses are looking underfed. Today Yvonne set off for Cressy to buy some hay, also the first time in years; if it hadn't been for that idiot Paul Ludovici last November, we'd have had plenty for ourselves and to sell.
Having to pay money for hay is only part of the issue: we needed somewhere to put it. It's true that the horses still haven't decided to use the new shed, but they certainly would if we put hay into it. Instead, we had to sacrifice part of the garage, and that in its turn meant that I finally had to tidy out the garage:
|
|||||||||||||||||||||||
Winter's nearly here, but we still have flowers:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The second tree (flower detail, then larger view) is covered in buds, and it has been for almost as long as we have been here. I wonder if the flowers will get more interesting when they all come into bloom.
| Thursday, 22 May 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More preparations for my presentation at Grain and Grape on Saturday. I have the slides ready (the paper won't be presented and thus hasn't been updated), but I need to bring a projector. Played around at length with eucla (formerly adelaide), my Inspiron 5100 laptop, attempting without success to get CRT switching to work. It displays both on internal and external screens in text mode (probably the BIOS sets it up that way), but X displays only on the external screen if it is present. Checked with a recent (2 month old) version of Knoppix, which wasn't able to switch either, though it did display X on both displays, which is good enough for me. It seems that the X driver has lost the ability to switch displays on this hardware.
But Knoppix can't read BSD file systems. Also tried tvremote, which displayed on both screens, but incorrectly—I had already noticed that several years ago. In the end, tried with Microsoft and found that it worked as intended. Rather sadly decided that I didn't really have time to install the newest version of FreeBSD—which I'm sure would work the way Knoppix does—so I'll end up using Microsoft to display on Saturday. Hopefully nobody will find out.
Also did a little bit of household accounting, including entering our rainfall into a database. In the last 9 months, since we started recording rainfall, we have had 542 mm rain, which should really be enough, but the distribution is a little uneven:
+-------+------+---------+ > Month | Year | sum(mm) | +-------+------+---------+ > 8 | 2007 | 14.9 | > 10 | 2007 | 19.7 | > 4 | 2008 | 23.2 | > 3 | 2008 | 30.6 | > 1 | 2008 | 34.4 | > 5 | 2008 | 52.0 | > 9 | 2007 | 55.7 | > 12 | 2007 | 79.6 | > 2 | 2008 | 95.8 | > 11 | 2007 | 142.3 | +-------+------+---------+
In particular, March and April of this year were very low, thus explaining our need to buy hay. Now I need to find a way to plot graphs which show the issues more carefully.
Also further discussion on IRC about the mystery plants we have. Came to the reasonably good conclusion that our mystery daisies are Osteospermum, particularly since I can find both the colours of the cultivars we have.
| Friday, 23 May 2008 | Dereel | Images for 23 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Up this morning to find some of the sprinklers dribbling, which they must have been doing for 10 hours—I had set the system to sprinkle at 22:00 to keep the soil moist in case of frost. Further investigation showed that the solenoid in question, the last I had installed, had failed. Damn! I hope the others last longer.
Almost as if out of solidarity, the house (rain)water system diminished to a trickle. Out to look at the pump—Davey, my favourite—and to puzzle out how it had been connected up: the output side (white in the photo) contains something like a filter, but I can't work out how to disconnect it:
|
|||||||||||||||||||||||
Instead undid a bolt on the input side, which proved to be a primitive mesh filter, full of conifer needles:
|
|||||||||||||||||||||||
Removed them and replaced the filter, and now we have water again, but it looks like I'll have to put something more substantial in the right place.
Why did people introduce a syntax for web mail? Instead of sending me a message to dontspamgrog@lemis.com, with a subject, say, I'll spam you anyway, people use the syntax mailto:dontspamgrog@lemis.com?subject=I'll%20%spam%20you%20anyway. Real MUAs don't understand that, and in any case, if it's in a web page, what can you do?
With the help of Callum Gibson (once again) discovered that there is something you can do. It's described in the comments of the script that I wrote for that purpose:
In firefox, go to the about:config page. Right click in the preferences and select New -> String.
Add a preference called network.protocol-handler.app.mailto.
Give it the name of this script (in this case, ~grog/src/startmutt), which must be executable and be in the PATH (yes, for once firefox respects the PATH).
And it works! But I wish it weren't necessary.
| Saturday, 24 May 2008 | Dereel –> Melbourne –> Dereel | Images for 24 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Up early this morning and off to Melbourne to Grain and Grape, arriving in plenty of time to find Nik dealing with a RIMS system he had never used before, which had somehow managed to get blocked wort lines:
|
|||||||||||||||||||||||
Clearly it won't solve all of my problems, but it's quite interesting, and I took a number of photos. Now I'll have to do some thinking about how to build my own. And RIMS or HERMS? It seems that the standard way of building a HERMS requires the water in the HLT to be the same temperature as the mash, which makes for a much slower ramp.
Then did my talk, which got a bit of interest. It's clear that the format isn't suitable for the environment (in the Grain and Grape storeroom). If I do it again I should consider how to change it.
Off then to Footscray market, where food was heavily discounted, in particular meat, but it was still more expensive than at the Queen Victoria Market, so didn't buy too much. Did find a grill plate for the grill we bought in Geelong last month, with the appropriate name “Rambo”.
Back home, and in the evening we had to inaugurate the Rambo, of course, and invited Chris over. Somehow Chris and Lilac have a special relationship:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I'm reminded of how the Germans discovered Sex.
| Sunday, 25 May 2008 | Dereel | Images for 25 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Chris Yeardley around this morning for breakfast—we had planned to finally go riding in Enfield State Forest, but strangely Chris didn't feel like it, and Yvonne wasn't overly keen on it either; one of the few times when it's not my fault that we don't go riding. This time we blamed it on the weather.
While Chris was there, she did discover an acacia tree with broken branches, which required a lot of work:
|
|||||||||||||||||||||||
The acacia is the tree at the left in the picture. In the process we found another plant, probably a eucalypt, which had been completely hidden by the acacia. Many of the plants in this garden were planted far too close to each other. We should trim the acacia back, but it's full of buds, and I'd like to leave it to bloom first.
That, of course, reminded me that I've been meaning to write a mystery plant page with photos of what we have here. That took a lot of work without being finished.
We have still more mushrooms! This time it's not difficult to identify them—they're Amanita muscaria:
|
|||||||||||||||||||||||
I wonder how they made it to Australia at all.
Why does exif truncate texts? I see something like this:
Date and Time (origi|2008:04:30 12:59:12 Date and Time (digit|2008:04:30 12:59:12 ComponentsConfigurat|Y Cb Cr - Exposure Bias |0.00 EV
Looking at the source code, there's no good reason; the texts come from libexif (the same package), but they has been arbitrarily limited to 20 characters. Changing it to 30 works, though that's still ugly:
=== root@dereel (/dev/ttypf) /usr/ports/graphics/exif 112 -> diff -wu work/exif-0.6.15/exif/actions.c~ work/exif-0.6.15/exif/actions.c
It's taken a while, but finally we have mice in the pantry again. There's no question that our best mouse trap is Lilac, who can catch a mouse in seconds if she has access. But we can't leave her in the pantry all night, so we dragged out a patent mouse trap. It's not good. A good mouse trap kills instantaneously by breaking the mouse's neck. This one is too slow, and on one occasion a mouse survived for 10 hours before Lilac relieved it of its misery.
Today we found another mouse in the trap. It was dead, but I'm wondering what it died of: the only place the trap hit it was on the tail. But it was so small that it fitted entirely into the trap. Maybe it suffocated:
|
|||||||||||||||||||||||
I must buy something better.
| Monday, 26 May 2008 | Dereel | Images for 26 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Into town today to get one of my carbon dioxide cylinders refilled. That turned out to be less simple than I thought: it seems that the cylinders need to be correctly identified and to have passed a hydrostatic test, and these ones haven't. It also needs a black identification stripe. That (mainly the test, of course) will cost me about $40—hopefully it'll pass—and I won't have the cylinder until Thursday.
Then to the Ballarat Pump Shop to attend to my house water problems. Why is it that there are so many fittings, and none of them are quite what I want? There are PVC pipes and two different kinds of “poly” (is that polypropylense?), not to mention “low density poly”, which doesn't really enter into this scenario. Then there are the metric sizes 13 mm, 19 mm, 25 mm and 32 mm, not quite the same as ½", ¾", 1" and 1¼", which are also available. The pump I have has a rural poly input, with the filter that I mentioned a few days ago, and the output goes into 1" poly, via a 1¼" non-return valve (according to some “experts”, they're important for Davey pumps, which can blow up otherwise) before going back to the 1" PVC again.
Connectors are another issue. I wanted to put in a filter, but the one I had was ¾", so I needed some adaptors, and of course there's the connection between “poly” and PVC (polyvinyl chloride). There are two different kinds of screw threads for poly, but you need cement for PVC. That's OK: I have cement, which I used for laying the power cables last month—but they need to dry for 12 hours before you can put water in them, which makes it rather difficult to use.
Finally took a collection of fittings to make two different ways of doing things, with the understanding that some of it would definitely come back. Also bumped into Matt, who did the installation of the bore pump, and who told me that my sprinkler solenoid was probably clogged, and how to fix it. Then on to Celsius to get some new solenoids, but they didn't have them—and the never will again: they're closing down. That's a nuisance, since they're the only place I knew with this particular part.
Back home to install the bits, and discovered that, despite all measurements and planning, they didn't fit the way I had intended: the adaptors made the line too long. Dragged out some odds and ends in the shed and ended up turning the pump through 90°; at least it works now. Only 3 hours work. Here “before” and “after” photos:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
CJ was also here today and helped Yvonne with the fencing. We're still not finished, but things are looking a lot better.
And the sprinkler? Turned it on, and it worked! One point to Matt; hopefully it'll stay that way. I still need another one, which I may end up having to buy on eBay.
| Tuesday, 27 May 2008 | Dereel | Images for 27 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Spent most of the day finally getting my gardening pages into useful shape, in particular the mystery plants page and offshoots. The result generated a fair amount of discussion on the #bugs (“Bringing Unusual Garden Specimens“) IRC channel, with the result, of course, that I had to go out and take more photos, notably of the succulent, the mystery flower of last Sunday, and of the “shade tree”:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
We've decided that the mystery flower is probably some kind of Oxalis
| Wednesday, 28 May 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Finished off the mystery garden web pages today, adding diary excerpts. That also involved a fair amount of PHP programming; my lack of understanding of the issues when I started is now coming back to bite me.
Another round of updating for “Beautiful Architecture” today, adding the images in DocBook format. That looks like:
<figure id="pic-1">
<title>Register stack</title>
<mediaobject>
<imageobject>
<imagedata fileref="pic1.pdf" format="PDF" />
</imageobject>
<caption>
<para>Discuss the figure</para>
</caption>
</mediaobject>
</figure>
For some reason, this works for JPEGs and GIFs, but XXE doesn't display the PDFs, though it also doesn't complain about them. It does completely reformat the source, mutilating it in the process.
Spent some time reformatting it and checked it back in again, and then read the editor's letter: I should run xmllint --postvalid against the file. Did that and got hundreds of errors, the simplest of which looked like:
grog.xml:1714: element para: validity error : Element para is not declared in para list of possible children
And that in a file that both XXE and nXML consider to be valid! Went back to the DocBook book and discovered no help at all; it's all about syntax, apart from the reference section. Finally gave up. O tempora! O mores!
| Thursday, 29 May 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The work I did on the garden diary yesterday started from 1 January 2008, and I had promised to add earlier data. That proved to be more difficult than I had expected, since the diary entries for 2007 hadn't been converted to PHP yet. Instead of doing something useful—the “Beautiful Architecture” chapter, for example—spent most of the day tidying up the main diary entries and then digesting them into the garden diary.
Why don't I store this stuff in some kind of database? That's the XML way of doing things, I suppose. But all such solutions impose constraints, and I suspect that most people who use them accept the constraints implicitly rather than explicitly. I'm completely flexible in the way I do it, admittedly at the cost of more work.
Finally got a look at the “Beautiful Architecture” chapter. The errors are fairly obvious once you see them. The problem is that the “tools” format the tags in a way that you can't easily compare opening and closing tags. XXE reformats valid text like this:
</itemizedlist> <para> When the backup process returns from <code>checkmonitor</code>, it has
become the new primary process. It returns to the location of the old primary's last call to
<code>checkpoint</code>, not to the location from which it was called. It then carries on
processing from this point. </para> <para> In general, the life of a process pair can like:
</para> <sect2>
<title>Synchronization</title>
HTML tidy has an option for reformatting XML, but it's not really much use:
</itemizedlist>
<para>When the backup process returns from
<code>checkmonitor</code>
, it has become the new primary process. It returns to the location of the old primary's last
call to
<code>checkpoint</code>
, not to the location from which it was called. It then carries on processing from this
point.</para>
<para>In general, the life of a process pair can like:</para>
<sect2>
<title>Synchronization</title>
It seems that most of my problems were attempts to nest paragraphs, which is not only syntactically incorrect, but also doesn't make much sense. I wish I could find a tool that reformats like this, which makes it clear where the paragraphs start and end:
</itemizedlist>
<para>
When the backup process returns from <code>checkmonitor</code>, it has become the
new primary process. It returns to the location of the old primary's last call to
<code>checkpoint</code>, not to the location from which it was called. It then
carries on processing from this point.
</para>
<para>
In general, the life of a process pair can like:
</para>
<sect2>
<title>Synchronization</title>
Unfortunately I haven't found one yet. nXML has something vaguely similar; I suppose I'll have to write a few ad-hoc Emacs macros.
And all the way through this article I missed a typo: “the life of a process pair can like” instead of “the life of a process pair can look like”
One of the things I haven't understood about the BST (Boosted Satellite Transmission) implementation on my satellite Internet connection is why it takes such a long time; connection setup and teardown is still in the order of 2 seconds. Today it seems to have had more than usual problems; despite “normal” IP and UDP traffic, and despite rebooting my modem, it took 3 hours to get enough bandwidth to download a mail archive. A real TCP connection would have timed out in that time. The graphs show only part of the pain:
| Friday, 30 May 2008 | Dereel | Images for 30 May 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
CJ back again today to help with the paddocks, and finally got things finished, including cleaning up a considerable amount of cut and fallen branches. Our hidden acacia tree is now visible:
|
|||||||||||||||||||||||
Finally Yvonne is happy:
|
|||||||||||||||||||||||
Chris and I have been thinking for some time about co-located servers, but for some reason every one we've seen so far has had problems. The biggest, clearly, is how far you can trust the ISP. The Germans seem to have good prices, but the stories I read in c't about ISP accounting practices worry me considerably. Also, I want a server that runs FreeBSD, and there aren't that many of them around. On the face of it, OVH look good—unlimited traffic, 100 Mb/s bandwidth, 250 GB disk, all for € 25 per month—but they require a German address. That wouldn't be the issue, except for my fears of the accounting practices I mentioned. Also, all feedback I have heard is negative.
But why Germany? From here it's two-thirds of the world away. The USA is the undisputed hub of the Internet, and there should be good suppliers there, only I haven't found one.
More tidying up the garden pages and also the garden itself. We have yet more mushrooms:
|
|||||||||||||||||||||||
Also added to the list of mystery plants:
|
|||||||||||||||||||||||
What is it? CJ's friend Sue thinks it's poisonous, but she doesn't know what it is. On the face of it it looks pretty enough, and though there's some growing in the paddocks, the horses don't go near it.
One thing that is becoming clear is that it can be worthwhile to procrastinate. I've taken a bit of a pause from digging up the entire garden, and now I discover that it doesn't have much effect on the amount of weeds. On the left is a dug up and planted area, and on the right one nearby that I had just sprayed a couple of times with glyphosate:
The plants on the left are Dimorphotheca ecklonis, previously called Osteospermum ecklonis. The title of the first photo is misleading; everything there was planted and intended to stay.
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Also more work on the DocBook stuff. Clearly the indentation is the key to many of my problems. Manually indented the document, placing the tags as I showed yesterday, and everything looked so much better that I was able to find my nested paragraphs and fix them. Peter Jeremy pointed me to a couple of XML pretty-printers in the Ports Collection, and I installed xmlprpr, but it did no better than HTML tidy. There are obviously two kinds of pretty-printing for XML: syntax-only and context-dependent. The syntax-only version, to which xmlprpr and tidy appear to belong, identifies tags but makes no distinction between them, putting each on a new line. This introduces white space which falsifies the document.
xmllint and XXE, by contrast, recognize the semantics (I suppose I should say “interpret the DTD”), so they can make a distinction. That doesn't make them good. Both of them seem to destroy the natural flow of the text, as shown yesterday. So it looks like we're still in need of better tools, a task not made any easier by the extreme complexity of the system. On IRC we were discussing which of the following statements contains more truth:
“XML: A solution in search of a problem”
“XML: A solution which is its own problem”
I like the reflexive nature of the second one.
| Saturday, 31 May 2008 | Dereel | Images for 31 May 2008 |
| Top of page | ||
| previous day |
It's been a year since we first saw our house, a good time to compare what's happened in the garden. I took a number of photos from the same angles as a year ago, but this one shows the most difference:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Presumably the difference in the daisies is due to the irrigation we installed.
What those photos don't show is the amount of work we were doing exactly in that area: it was far too densely planted—for example, I found a rose planted 30 cm from one of the daisy bushes. Trimmed back the hibiscus and removed two sorry-looking roses; had to cut through the roots, so they'll probably die, but replanted them anyway to give them a chance. Yvonne tackled the lilac (the bare tree in the middle of the photo), which really needs a lot of trimming back, but I'm not sure how best to do it. That took us most of the day.
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 |
