|
| Greg's diary |
| February 2012 |
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
This view of the diary is limited to these topics: technology. There may be lack of continuity in the text, and some days may be completely missing. In case of doubt, please enable the complete display.
| Wednesday, 1 February 2012 | Dereel | |
| Top of page | ||
| next day | ||
| last day |
|
Nickel-Zinc battery reliability
|
Topic: technology, opinion | Link here |
I've been quite happy with the Nickel-Zinc batteries I bought a few months back, and so I bought some more. They arrived yesterday: 8 AA size and 4 AAA size. It took me a couple of days to charge them: that's 4 loads (for some reason the charger handles only 2 AAA batteries at a time), and each takes 5 hours.
And one AAA battery didn't charge properly. After the charge cycle was complete, one battery had 1.836 V, which is about normal, and the other only 1.699 V, which is definitely too low. In the course of time that dropped to 1.368 V. A second time round it didn't charge at all, but later I got it to go as high as 1.7 V, but after a few hours it fell back to 1.62 V. Clearly it's defective.
It's always possible that a battery is defective, of course, and as long as it's only one, it's not even statistically relevant. But it begs the question: how reliable are NiZn batteries? And more generally, how reliable are other kinds of battery? I've already spent some time monitoring the charge of my Lithium-Ion batteries for my camera. Time to do something more general.
| Friday, 3 February 2012 | Dereel | Images for 3 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
The spammer's apprentice
|
Topic: technology, opinion | Link here |
Received another unlikely looking spam today:
Does nothing get tested nowadays?
| Sunday, 5 February 2012 | Dereel | Images for 5 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
| Tuesday, 7 February 2012 | Dereel | Images for 7 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Still looking for a keyboard
|
Topic: technology, opinion | Link here |
My current keyboard is a Northgate OmniKey keyboard manufactured in August 1989, 22½ years ago. It's no longer in the best of condition, but in all that time I haven't found anything that I would like to replace it with. Things are getting desperate, though. The r key, in particular, is bouncing badly.
But it's not the only keyboard of that kind that I have, and in the past I've found that if I rotate between them, the bounce tends to recover. So today I went to see what I could find: two Avant Stellar keyboards, both with defective keys, and three other OmniKeys (I thought I had five, but I can't find the fifth), all with their own problems. Tried them in turn and found: one had a defective Control key and a couple of flaky keys; another had really severe bounce in the Space key; the third had a broken clip on the Enter key.
With a bit of chopping and changing (moving parts from one keyboard to another) I can probably get a few more years' life out of them. But as I've commented before, this is stupid. Why can't I find a modern keyboard that will satisfy me? There's also the issue that PS/2 keyboards are obsolete, and the modern replacement is (shudder) USB.
So what do I need from a keyboard? The keys should feel “right”, of course, but most modern keyboards, even the cheapest, aren't too bad in that respect. But they're clearly designed by people who can't touch-type. In particular, about 25 years ago the function keys were moved to where you can no longer find them without looking at the keyboard. For some reason, they've also done it with the Esc key. That's the biggest issue I have. Others, such as the location of specific keys (`/~ and the Ctrl and Alt keys) depend on the use of the keyboard, and they can be relatively easily remapped (except for those brain-dead keyboards that hard-wire the completely useless CapsLock key). And though the “Inverted T” arrow keypad layout makes no sense at all, it's not really that painful, though like any keypad it requires moving the hand from the main keyboard.
So: the big issue is to find a keyboard with function keys on the left, where I can access them without taking my eyes off whatever else I'm looking at. They must be available. Time to start looking.
| Friday, 10 February 2012 | Dereel | Images for 10 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Finding a new keyboard
|
Topic: technology, opinion | Link here |
Spent some more time looking for replacement keyboards today. The one must is a function key block on the left—as I discovered in Wikipedia, others agree with me:
Early models of Enhanced keyboard (notably those manufactured by Northgate Ltd.) maintained the layout with function keys on the left side, arranged in two columns of six pairs. This layout was more efficient for touch typists but was superseded in the marketplace by that with F-keys along the top.
But where can I find one? I recall having found a number of links a while back, but I can't find them any more. Looking on eBay and similar didn't help much. And it's not as if I can buy just any keyboard with F keys on the left, either: I already have a very dubious one which I refuse to use.
But then I have at least two other keyboards which almost fit the bill: Sun Type 5c and Sun Type 6. The main problem is that they're electrically incompatible. Did more investigation and came up with a number of pages describing how to build USB adapters: http://kentie.net/article/sunkbd/ describes what appears to be a prototype to adapt a Type 6 keyboard, but it requires a lot of work, including programming microcontrollers. http://gis.sakura.ne.jp/sunkey/sunkey_e.html describes a similar one for a Type 5 keyboard, but it doesn't even have a circuit. http://www.buzzard.me.uk/jonathan/sunkbd.html, also for a Type 5 (is there any difference?) gives much more detail, even a PCB mask.
But again it requires a microcontroller. As a kernel hacker, I'd rather minimalist hardware and a kernel driver. The first link above provides some Sun documentation for the keyboard protocols, which show it to be 1200 bps at TTL levels. It should be relatively trivial to connect it to a serial port, or a serial/USB adapter. But I'd much rather find something that doesn't need any custom hardware at all. More searching needed.
Of course, even that may be overkill. The Type 7 keyboard has a USB interface, and they're available. But I've found various reports on the web suggesting that people have had trouble with them, I don't have one, and though they're relatively cheap second-hand (from $29), they're all in the USA, and they want more than double that price for shipping. And what happens then if I find that I don't like it? So for the time being I'll investigate the option of a TTL to RS-232 level adapter.
| Saturday, 11 February 2012 | Dereel → Ballarat → Dereel | Images for 11 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More keyboard ideas
|
Topic: technology, opinion | Link here |
Mail from Ian Donaldson about my keyboard problems, suggesting I look at the Northgate “Ultra Plus”. I suppose the combination of the length of this thread and the description have obfuscated things. He found the information at http://www.northgate-keyboard-repair.com/, where the keyboard is really called “Ultra Plus”. But of course it's an OmniKey Plus, and it's identical to what I'm trying to replace:
But he has a point. He suggested buying a new (i.e. different, reconditioned) one, but I don't need that. I can repair the existing ones. It seems ridiculous, but I have all the equipment I need, and it's less fiddly than building PCBs with microcontrollers. I might have a go at it.
| Sunday, 12 February 2012 | Dereel | Images for 12 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Setting names for photos
|
Topic: photography, technology, opinion | Link here |
Chris Yeardley has been on holiday in Việt Nam, and I'm helping her bring her photos on line. One of the (many) reasons I don't do this with commercially available software is that I want to give titles to my photos, not names like 100_4984.JPG or even (shudder) 36836231@N00/2423055893. Yes, you can do that. I've tried it with various products, and you'd think that they deliberately make it as complicated as possible.
My current method works much faster: a little script that displays image name and details and prompts for a name. It uses GNU readline, so I can copy the name I apply to one image and apply it to the next with a different extension. Typically I'll have names like Pelargonium-1, Pelargonium-2 and Pelargonium-3, so I can copy the text Pelargonium- and just add a digit. Independently of this, I use xv to display the images.
Still, it's not easy to remember what you've done, in particular the sequence numbers. This is most evident in my monthly flower photos, where names such as the example above may be spread around through a list of 130 odd photos. And Chris ran into this problem today as well, exacerbated by the fact that she only had a single screen, so the xv image display overlaps the xterm window with the script. It was interesting to watch Chris use it—she's the first person apart from myself ever to do so—and it became clear that something better was necessary.
But what? I already have a script that builds web pages with “contact prints”, which simply shows all the images reduced on one page. But it should be possible to write code that accepts text input, sets default names, automatically completes the pelargoniums and checks for duplicates.
But how? I suppose the correct answer is “JavaScript”, but I don't know that well enough. Straight HTML with PHP should also do the trick—but how do I get a page to submit the contents of all forms? My first approach looks like this:
|
||||||||||||||||||||||
Each input field is associated with a photo, so it can't be in a single form. How do you do that? Can you do that? None of the documentation I found, either on paper or on the web, tells me that. So, once again, it's a lot of guesswork and experimentation.
|
ALDI GPS receiver: a step backward?
|
Topic: technology, general, opinion | Link here |
Took a first look at the GPS receiver I bought from ALDI yesterday. It uses different software, and so far I'm unimpressed. The maps are as bad as ever. The street where I live is still not there (it's only been here for about 100 years), and the fantasy streets that have been added to the east in the last couple of years are still there:
|
|||||||||||||||||||||||
My house is in Kleins Road, roughly on the l of the “Please tear off” foil. It's shown as less important than the (non-existent) road going off to the bottom through the arrow of the More tab. Finding addresses is now an order of magnitude worse. Previous versions allowed you to select “City Centre” or “Street middlepoint”, but now it seems that you must select one of the streets it knows and a street number that it knows in that street.
Admittedly, I haven't checked the manual—in my experience, that wouldn't help—but the fact alone that the maps are still broken means that it will go back. The rest just makes it even worse.
| Monday, 13 February 2012 | Dereel | Images for 13 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Contact prints page
|
Topic: technology, photography | Link here |
Got various feedback from people reading this diary about how to make a single form to include all my <input> tags. The one I looked at, from Peter Jeremy, included <span> tags inside the form. That appears to work, but according to the definitions at http://www.w3schools.com/html5/tag_form.asp it shouldn't:
The <form> element can contain one or more of the following form elements:
This is for HTML 5, being the latest and greatest, but things aren't significantly different for other versions. You can't put general content inside a form. This appears to be Yet Another Example of where the HTML syntax is unnecessarily restrictive. What I'm trying to do is pretty much what almost any Tandem Pathway program would have done 30 years ago. Why do I have to break the rules to do it?
Spent some time playing around with that, iteratively improving things, and doing other things in between. It's getting there, but it's a long way from finished, and it grates that I'm breaking the rules.
| Tuesday, 14 February 2012 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More contact prints and HTML insights
|
Topic: technology, opinion, photography | Link here |
Yesterday's diary prompted more response. In particular, Edwin Groothuis pointed at the HTML 4 specification, which says:
An HTML form is a section of a document containing normal content, markup, special elements called controls ...
That's not quite the same as what the w3schools document, but on re-reading that document, it doesn't make it clear that the listed elements are the only ones allowed. Still, that's HTML 4, and one day we'll have HTML5 (and save spaces in the process) What does that say?
A form is a component of a Web page that has form controls, such as text fields, buttons, checkboxes, range controls, or color pickers. A user can interact with such a form, providing data that can then be sent to the server for further processing (e.g. returning the results of a search or calculation). No client-side scripting is needed in many cases, though an API is available so that scripts can augment the user experience or use forms for purposes other than submitting data to a server.
That's about as vague as the w3schools document, and it introduces more bad language: abuse of the word “experience” to mean “impression”. I suppose language has to evolve, but why entrust it to buzzword marketeers? Does it have to eclipse existing usage? Or are we to suppose that there are no experienced people left? There's also a term “non-normative” whose meaning is not immediately evident.
The rest of the HTML5 document shows that yes, indeed, I can do exactly what I want with the <form>, so I did. With a bit of work, it did what I told it. Now I need to think about how to make it easy to use. One clue might be in that “user experience” comment: do it in JavaScript. But that requires more experience than I have.
| Wednesday, 15 February 2012 | Dereel | Images for 15 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Yet Another Useless Web Site
|
Topic: technology, opinion | Link here |
While in town, to the Ballarat Library, which had just informed me that the DVD that I had returned on Saturday was available for pickup. So was “South Pacific”, so I picked up that, and then discovered it was a BBC Documentary, not the musical I was looking for. This web site is really impossible! Last time I had spoken to them about it, they sang the praises of how good it was, so this time I asked to be shown how to do it and was helped by a Senior Librarian. First question: “What is the PIN number of your library card?”. It wasn't necessary, and I didn't give it. But I don't suppose it'll stop her asking somebody else next time. She showed me a couple of non-obvious tricks, but she still wasn't able to perform some normal searches. “Java? Is that a computer language?” At least I made my point, and she told me they're considering a replacement web site. Hopefully I was able to give her some useful input.
|
Photo naming page: success
|
Topic: photography, technology | Link here |
Today was the first day I used my new photo processing web page in earnest. It's still pretty bare-bones, but it includes one feature that makes life easier: a single , (comma) in a field tells the page to take the previous name (less trailing number) and append the next sequential number: after the image Bunnings-blockade-1, a comma will generate the name Bunnings-blockade-2. I can already see the use of a ! to say “continue doing this until you find another specification”. And I can see myself playing with this for some time to come. PHP is much more malleable than shell scripts.
| Thursday, 16 February 2012 | Dereel | Images for 16 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
But none came. Could the policeman have lost the details? In any case, we weren't too upset. But yesterday a reminder arrived, by some stroke of bad luck. Based on the address, it should never have got here:
I suppose it's typical of the attitude of these people that they've turned the name around without indication. What if it were for my friend Olufemi Adekola? Or is that Adekola Olufemi? Of course, it doesn't have to be a non-European name to be confusing. In Germany 30 years ago I had neighbours called Peter Phillip and Otto Adolf.
But Enfield? We live in Dereel. Yes, there's a hamlet called Enfield 10 km away, but what does that have to do with us?
How can that happen? Apart from a really draconian fine, they want to put 3 demerit points in her driver license. And for that they need (and have) details of her driver license, where the address is almost correct:
And there, again, the address is wrong! The house number is 47. These details are important. How can the authorities make this kind of mistake not once, but twice? This kind of incompetence stands out from minor lapses keeping to an imposed speed limit inappropriate to the road conditions. In this case, though, Yvonne reported the error long ago, and they put a sticker with the correct address on the other side.
Things wouldn't be so bad, except that they wanted $20 more to pay for the delay caused by their incompetence. Yvonne called them up and was told to write a letter explaining the problem. Given the fact that they had all the evidence in front of them (address to which the summons was sent, address on the driver license). But no, Yvonne had to write a letter. She did so, handwritten, with the request that they drop the excess charges (which still seems to be a matter that they can decline), and asked for an explanation about how this could happen in the first place. I don't have much expectation that she'll get that—but why shouldn't she? The authorities are supposed to be for the public good, so they should be accountable.
End deselected topic ho -->|
Limits of DxO Optics “Pro”
|
Topic: photography, technology, opinion | Link here |
I've found that when processing garden photos with DxO Optics "Pro", they generally come out best with the “HDR Artistic” “preset”. There are limits, though, apparently when the contrast is very low. Here a comparison of one of yesterday's photos done with the “Artistic” and “Realistic” presets:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The artefact at the bottom on the first image was generated by the software.
|
Another crash
|
Topic: technology | Link here |
dereel crashed today, out of the blue. There used to be a time when computer crashes seem to happen like this all the time, for no particular reason. But while that may still be the case for Microsoft-based machines (I really don't know), it's now very seldom on any of my BSD machines. I wonder if it's an indication of hardware problems (something that people love to blame software issues on).
The thing that did get me was the time it took to fsck my photo disk. OK, it's 2 “TB”, of which 60% are in use, and it has over 400,000 files on it:
It took 2 hours, 40 minutes, mainly in phases 1 (2 hours) and 5 (30 minutes). Phase 5 stuck at 92% for 27 minutes of that time. I wonder why. One thing that did hit me was the slow I/O transfer rate, only about 40 to 50 transfers a second in phase 5. Is that because of extreme seek distances? I suppose it's time to read the source.
| Saturday, 18 February 2012 | Dereel | Images for 18 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More Nickel-Zinc experiences
|
Topic: technology, photography, opinion | Link here |
Changed the NiZn batteries in my flash gun today, and recharged them. Nothing unusual in that—almost. I had forgotten to turn the flash gun off last week, and though it shuts down automatically after a while, it's a soft shutdown (it has a real mechanical power switch). But the voltages were interesting: before recharge, 3 of them were in a normal range, and the fourth was so low that the standards say it's defective. But they all recharged happily:
| Battery | Charge date | Before | After | |||
| 9 | 20120218 | 1.695 | 1.866 | |||
| 10 | 20120218 | 0.789 | 1.845 | |||
| 11 | 20120218 | 1.682 | 1.884 | |||
| 13 | 20120218 | 1.659 | 1.875 | |||
Yes, battery 10 had the lowest voltage after charging, but only barely. It'll be interesting to keep my eyes on it.
| Sunday, 19 February 2012 | Dereel | Images for 19 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More photo processing
|
Topic: photography, technology | Link here |
Chris Yeardley over this morning for a Nasi Lemak breakfast, and to continue processing her photos with my new, super-duper web forms. She brought a book on JavaScript with her. I suppose the best thing we can say is that she found my software better than the previous attempt. In the end she finished naming the things and left the conversion to me—not unreasonable, considering that my software is not just undocumented, but under active development. It was also interesting to see how few assumptions I had made that the software was in my directory hierarchy, though I needed to make a fair amount of relatively trivial changes.
|
Garden photos with new software
|
Topic: photography, gardening, technology | Link here |
While Chris was doing her thing, I took my monthly garden photos. They're a particularly good test for the new software, since there are about 150 of them, and they have names like Petunia-1 and Petunia-2, but those names are interspersed with other images. Managed to get that sorted out, and also used my first-ever HTML5 feature: the autofocus attribute to tell the browser where to position the cursor. All worked surprisingly well, but it also makes clear that I'm like the man who only has a hammer: every problem looks like a nail. I really need to convert this stuff to JavaScript.
| Monday, 20 February 2012 | Dereel | Images for 20 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Ballarat Library: another false positive
|
Topic: general, technology, opinion | Link here |
While in town, also dropped in at the Ballarat Library to pick up a couple of DVDs I had requested. One was “The Italian Job”, coincidentally (in 1969) one of the first films involving computer-related crime. The web site couldn't find one in the Ballarat library, but that didn't stop it offering me 13 irrelevant titles. Finally I found one, and of course it was wrong again: it wasn't fiction. That's two out of three DVDs in the last couple of weeks that, thanks to inadequate documentation. Now that I have a few more tricks up my sleeve, took another look (“search all libraries”). 10 titles, all of them either “DVD” or “Videorecording”. Some of them even have dates: 2002, 2003, 2004. It proves that there was a remake done in 2003, and that seems to be more popular (at least with librarians) than the original 1969 version. But what's 2002 and 2004? After much searching, discovered that the 2004 version was the 2003 version, and the 2002 version (available in Myrtleford and Mildura) is the 1969 version. So I put a “hold” and got this information, which makes it immediately clear which one I am going to get:
Hold(s) Placed
The Italian job [videorecording]
DVD F ITAL
Pickup at: Ballarat Library
Looking at the HTML is even more revealing. It had many empty lines, which I've removed here:
I particularly like the colspan specification.
|
Finishing Chris' photos
|
Topic: photography, technology, opinion | Link here |
Spent much of the afternoon working on Chris Yeardley's photos. In principle there's no hurry, but it required significant uploads—there were a total of 575 MB of photos, which also overflowed the tiny disk we have available on the server—and today was the last day of our monthly ISP billing period. We had 1.2 GB (out of 9 GB) over, so it was clearly the day to do things.
Lots of experiences in the process, mainly related to resetting the camera times. I've seen this before with the Hackers barbecue two years ago, and I wrote a program to do that at the time. But it's still amazingly complex, and it's far too easy to do it twice or in the wrong direction. Apart from that, I found myself revisiting code I wrote years ago, and also installing my own infrastructure for her. Nothing really difficult, but somehow there were so many things going on today.
We certainly used up our quota. Today's also traditionally the day where Yvonne soaks up our remaining allowance with YouTube downloads, and we ended up overrunning our quota by a small margin:
| Day | Download | Upload | Total | |||
| 20/02/12 | 469.7 | 878.0 | 1347.8 | |||
| Above Period Total | 4043.7 | 5100.9 | 9144.5 | |||
| Tuesday, 21 February 2012 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
wview: Mine is better than yours
|
Topic: technology, opinion | Link here |
About 2½ years ago I bought a Fine Offset WH-1081 weather station, which of course came with software only for Microsoft, and so even before I bought it, I had to find software that could talk to it.
That was a set of patches to an old version of wview that ran only on NetBSD, an operating system that I no longer have running all the time, so I had a fair amount of work to do: apply the patches, migrate from NetBSD to FreeBSD, and update the patches to the latest version of wview.
I never finished. Somewhere along the line I gave up and wrote my own software instead. Why? wview is a relatively complete system, but it gave me so much pain that I thought it would be easier to write my own software. I was right.
It seems that the author of wview has recently become aware of something I said, though he doesn't say what. Today I received a message from him that shows that the best tradition of the USENET flame is not yet dead:
Finally, your entire attitude about other people's work truly is emetic. You are either becoming senile (and thus a moron) or you are just a moron.
In passing, this use of the word “emetic” is interesting. I use it a lot, but few others do. On asking, it proved that he had, indeed, copied my usage. To quote Charles Caleb Colton, “Imitation is the sincerest form of flattery”.
Still, he raises some questions. Why did I dump wview? It's a mature system with many features. I didn't do it lightly. There are a number of reasons. Looking back at my diary of the time, I find:
It didn't support my weather station. I knew this, of course, but it meant significant effort just to get it to work with the station.
It used GNU autotools, arguably some of the untidiest build tools I have ever seen, and one that I couldn't get to work correctly.
Error reporting was flaky. It had a requirement that passwords in the configuration file must not have spaces in them, but it didn't complain if you put them in; it just truncated them, causing me significant problems on at least one occasion.
It used the Three Ugly Sisters, System V IPC (SHM, SEM and MSG), or at least System V semaphores, something so distasteful that nobody appears to have written a Wikipedia page on the subject. They also caused problems when the processes crashed and didn't remove the old data structures.
Starting wview had to be done with a script, and it was very difficult to ensure that the initial conditions were right (no processes running, SEM semaphores removed). It also meant that I couldn't find a way to debug the daemon, I think because of timing issues with other processes.
The information returned by the station is in metric units, sort of. The wind speed, for example, is in units of 0.1 m/s. wview translated everything into American units first, and if you selected metric units it translated them back again. In the process it lost a lot of accuracy. 0.1 m/s is 0.36 km/h; wview stored in units of 1 mph (1.6 km/h), so it was returning values like 2, 3 and 5 km/h, but not 1 or 4.
Though wview had many features, there are some that it didn't have, including comparisons between weather stations and loading archival data from the station.
The interpersonal communication with the project author proved to be abrasive. In particular, he required that I reply to mail messages upside-down. I don't do that, so I couldn't communicate. Today's message suggests that nothing has changed there.
There are other issues, like hard-coded pathnames such as /usr/local, which made it a pain with NetBSD, and two different databases (sqlite3 for the configuration and MySQL or PostgreSQL for the weather station data), but they're just uglinesses, not enough reason to drop the product.
So what options were left? Modify wview to do what I want? Yes, that would have been a possibility, but for that it's too ugly. By the time I came to the conclusion to ditch it, I had spent over two months work. Despite significant pain with libusb, I had my own version up and stumbling in over a week, and it had replaced wview in two. It would be incorrect to say that it has worked to my satisfaction over the last more than two years, but I've never had the pain that I had with wview.
What would I do now? wview now supports the my weather station. Maybe it's more stable. But it still won't do some of the things that I want. So I'll compare the code carefully and possibly come up with some improvements in the protocol, which is still flaky.
There's more to say, but I'll wait for the results of further correspondence with the author.
| Thursday, 23 February 2012 | Dereel | Images for 23 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Tracking the DxO bug
|
Topic: photography, technology | Link here |
Right from the beginning using DxO Optics "Pro" I had a problem that it didn't preserve the EXIF data from the original images. Not that big a deal: I wrote a little script to copy the data from the source. But after all, this is commercial software, and I am entitled to support, so I sent in a problem report and got a very quick response: “It works for me”. That wasn't a “case closed” situation, though: they gave me the source of an image to compare with, and a lot more details of how to report the problem.
Today I (finally) got round to downloading the image, which wasn't easy. It took me about 4 hours and 18 attempts to get it; the connection kept timing out. Tried the image, and how about that! It worked. Also tried some comparable images with Chris Yeardley's Nikon D70S, and that worked too. So what's the issue? To be thought about.
| Friday, 24 February 2012 | Dereel | Images for 24 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
DxO bug: found?
|
Topic: photography, technology | Link here |
Yesterday's experiments showed that the EXIF data problem with DxO Optics "Pro" was limited to my images from my Olympus E-30. Even DxO's own images didn't trigger the problem. But why? What's the difference? Different firmware? That's hardly likely to change the image file format.
Then it occurred to me: for each raw Olympus file that I read in, I perform (effectively):
Could it be that? Tried adding an author entry to DxO's trial image and bingo! it (silently) didn't copy the EXIF data. Tried it on one of the raw images from Chris Yeardley's Nikon, and things still worked correctly. So it's an issue with Olympus and custom EXIF data. Now to prepare a report, including—as requested—logs of what I've done.
|
ALDI “Set-top box“
|
Topic: multimedia, technology, opinion | Link here |
The third tuner on cvr2, my TV receiver box, has died. Yes, I can buy a new one, but a number of reasons speak for getting a “set-top box”, a modern word for “tuner”: the price isn't significantly different, you can use it to watch TV directly, it's a separate piece of equipment, so it probably won't fail at the same time, and it's kosher when calling up TV stations reporting reception problems: in the past when I've reported problems, they asked me what equipment I had, and when I told them it was a computer, they weren't interested in investigating. ALDI had one on offer a couple of weeks ago, so I bought it, and today (first time I needed 3 tuners running at the same time) I set it up. It has a USB slot for mass storage, so I dragged out an old random 8 GB USB stick and tried it out. “USB device inserted”. “No disk found”. I guessed—correctly—that the latter message meant “what kind of partition is that?”.
So: create a FAT32 file system on the
stick. How do I do that? I have enough pain with FreeBSD USB interfaces that I didn't even try. Put it in boskoop, my Apple,
instead. “Partition type not recognized. Initialize?”. OK, tried that. After 5 minutes
of “erasing”, while iostat showed about 40 single-sector I/Os a second, I gave up.
No way to cancel, so I just pulled the stick, and after a while
the programApplication stopped with no comment.
Put it into braindeath, the Microsoft XP box I have on loan from Chris Yeardley. It told me I had to reboot before it could do anything with the stick. I thought those days were over. So I put it in dereel, my FreeBSD box, and ran mformat, which was happy enough, though it, too, only managed 40 I/Os a second, so maybe that's a function of the stick. But it finished in a fraction of the time I waited on the Apple. I wonder if Apple really tries to write the entire device.
Once in the tuner, it was relatively simple to record a programme. And I could take it out and plug it into teevee, the TV player computer, and play it. Almost. This stick is so slow that I couldn't seek, and there were reception problems that I didn't have on a simultaneous recording on cvr2. They could really be reception problems, but at that speed it could just be data overrun.
All in all, though, not a bad result. Now I need to find a spare USB disk and try again with something that can handle the data rates.
| Monday, 27 February 2012 | Dereel → Ballarat → Dereel | Images for 27 February 2012 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
New web site with old computer documentation
|
Topic: technology, history | Link here |
While looking for information about the UNIVAC 1108 Master File Directory today, stumbled across bitsavers.org today. It's not new, and—most emphatically—neither is the content. It includes documents that were old when I entered the industry 40 years ago. It looks as if there's a lot to explore there.
My interest in the MFD was a supposition that it might have a relationship to the Unix directory structure. It doesn't. It is much more complicated, centralized, and has features that are rare today, such as file versioning at a basic level. That in itself is worth reading up on, on page 537 of the “programmer's bible”, UP 4144. This is the same manual that I had put on my desk on the first day of my working life, at UNIVAC in Frankfurt/Main on 2 May 1973.
| Top of page | Previous month | Greg's home page | Today's diary entry | Next month | Greg's photos | Copyright information |
