For more topics, scroll down,
press Ctrl-F to search the page,
or check previous months.
For the latest edition of this page at any time,
use this link: www.covingtoninnovations.com/michael/blog
2016 August 31
If I ever have a heart attack or stroke, squeeze my arm
An interesting medical discovery that I've been following is
remote ischemic (pre)conditioning.
It has been discovered that if any large muscle is deprived of blood flow, it sends out
some kind of chemical signal telling the whole body to prepare for the same thing,
and this signal reduces the amount of damage done by heart attacks and strokes.
It can be done before, during, and/or after the heart attack or stroke, any time before
the damage is complete, and it will reduce damage.
Simply inflating a blood pressure cuff on an arm 4 times is enough to make a measurable
difference, according to this study,
which confirms earlier ones. (The University library has the whole article on line,
but only people on campus or logged in can use the link.)
This hasn't made it into standard paramedic/EMT practice yet, but it ought to.
The best thing about it is that as far as they can tell, no one at all was harmed
by doing it. It seems to be the safest procedure in all of emergency medicine.
In fact, if I ever undergo a procedure where a heart attack or stroke is a risk
(such as cardiac catheterization), I'm going to specifically ask for remote ischemic
conditioning. It's quick, easy, and safe, and it improves the odds, and if you don't
need it, it doesn't have any effect.
This isn't a great picture of the Ring Nebula (M57), but the weather has made
astrophotography almost impossible most of the time for weeks, and these days,
any astrophoto is better than none.
I'm still testing a friend's AT65EDQ 65-mm f/6.5 apochromatic refractor.
This time I put it on the AVX mount and
piggybacked my STV autoguider with a 35-mm binocular objective.
Since this was just an equipment test, I didn't take flats, flat darks, or dark frames.
(Nor do I seem to have needed them!)
What you see here is a stack of six 3-minute exposures.
You're looking at the center of the picture, at the full resolution of
the Nikon D5300, about 1.9 arc-seconds per pixel.
I think it's safe to say the guiding and the optics are OK!
On the evening of the 27th, Venus and Jupiter were close enough together in the sky
to fit into a single telescope field. I chose to photograph them with a 300-mm
telephoto lens. To get a good west horizon, I went to a UGA parking lot on Lumpkin Street,
a former fraternity house site, overlooking the recently-torn-down southern part of
the Holiday Inn.
Venus is the upper one in the picture, with some diffraction spikes due to its
brightness. Jupiter is the lower one, more or less correctly exposed. Venus actually
appeared closer to Jupiter than Jupiter's satellite Callisto (not visible in the picture).
Of course, Jupiter is much farther away in space than Venus; this is just an accidental
alignment.
Important note: This problem was apparently fixed by update KB3194496 on September 29.
More information here.
I've pieced together a little more information about Microsoft's webcam blunder.
It is not just (as some media report) that they dropped some compressed video formats.
More importantly, in order to enable Cortana and some of their other apps to use the
webcam simultaneously, they added another layer to the OS between the webcam
and the software that's using it. Software no longer has low-level control of the camera.
Someone pointed out that this is a security risk. Now there is a place where a virus
could intercept your webcam images, and send them to the bad guys, without disrupting
the software you're using with the webcam.
The immediate effect is that by removing low-level camera control, Microsoft has
exited the scientific imaging market. In the long run, new camera drivers will
presumably restore low-level camera control, but we are losing the biggest advantage of Windows
for such applications — that it made webcam programming easy.
Microsoft, please just back out of this error!
[Update:] I am getting occasional reports that some astronomical cameras (other than my DFK
and DMK) work fine (with some, or all, PC hardware; I can't see how the PC hardware makes
a difference). Apparently, some drivers do not use the functionality that Microsoft has blocked.
Spurred by the way Microsoft let me down, I not only rolled my astrophotography laptop
back to Windows 10 1511, but also gave it a Linux partition and set it up to do planetary
imaging under Linux. The available software is much less abundant and mature, but
oaCapture
seems to be meeting my needs with DFK and DMK cameras.
Due to our cloudy weather, I haven't done anything more than dry runs inside the house;
I don't have any Linux-based astrophotos yet. But here's what I know about how to get
started.
To paraphrase Mrs. Beeton, First, catch your Linux.
The Linux subsystem in Windows 10 1607 is no help because it doesn't use Linux
device drivers.
You need actual Linux Mint (or Ubuntu or whatever you prefer).
I went with Linux Mint 17 rather than 18, since I like things that are tried and true.
Download the ISO file and make yourself a Linux boot CD.
You can easily set up your computer to dual-boot Windows 10 and Linux, but you have to do slightly
more preparation under Windows than in the past.
Here's how.
Basically, you use Disk Manager to free up at least 20 and preferably 100 GB of unallocated space, and
then you install Linux into it by booting from the Linux CD.
All of this is familiar to me since I've been using Linux occasionally for other purposes.
Next, download the appropriate oaCapture executable.
(Later on, oaCapture will be in the package distribution system, but not yet.)
Unpack the folder (in Linux Mint, just save it on the desktop, right-click, and "extract to here").
Open the folder with Terminal and install:
sudo ./install-binaries.sh
Also (and this is important) add yourself to the "users" group; only members of that group
can use the video cameras:
sudo addgroup username users (using your user name)
Now you can start oaCapture from Terminal by typing oacapture (lowercase).
To add it to the menus and desktop, right-click on the desktop, create a launcher,
and make it refer to /usr/local/bin/oacapture.
Now you're all set. If you have a small screen, as I do, you'll probably want to go to
its Settings, General menu, choose "Make controls dockable," and move the sub-windows
for the controls into a convenient arrangement.
Important note: This problem was apparently fixed by update KB3194496 on September 29.
More information here.
Microsoft is apologizing for forgetting to tell us (and webcam and software manufacturers)
that some kinds of webcam support would be dropped in Windows 10 1607.
Having let us down, I think they have an immediate ethical obligation to
allow everyone who upgraded to 1607 to drop back to 1511,
whether or not they are still within the 10-day rollback period, and
whether or not they possess the recovery files.
That means making a release of 1511 that will install over 1607 like an upgrade.
Perhaps such a release already exists.
Either that, or they must immediately release an update that completely undoes
the breaking changes, even if it breaks the latest version of Cortana or some of
their other smartphone-like toys.
We buy operating systems to run software and hardware, not to play with them.
Whichever of these they choose, we need it today, not next month.
Microsoft risks losing the trust of scientific and audiovisual users forever
if they do not act quickly.
If you agree, feel free to share links to this entry.
The computer I use with my astronomy cameras is an Asus UX32A Zenbook.
It actually had three problems that began with Windows 10 Anniversary Update
(build 1607) or possibly some update that came shortly afterward:
The astronomy cameras (webcams) didn't work, even with the latest drivers.
The touchpad was excessively sensitive, and no way to adjust it was provided.
The computer seemed slow overall; I only noticed this after it sped up after the rollback.
Conclusion: Windows 10 Anniversary Update was not ready for prime time.
It wasn't adequately tested.
It's very rare for an update to introduce driver problems, and I don't fault
the hardware manufacturers for not making new drivers. Nobody told them to!
This is still Windows 10, and nobody knew Microsoft was going to change things,
under the hood, that would interfere with hardware.
Maybe new drivers will come out
that work around the problem; maybe Microsoft will fix it. I'll know when the
cameras start working again with my Lenovo laptop, which I keep fully updated.
Important note: This problem was apparently fixed by update KB3194496 on September 29.
More information here.
Anyhow, here's how I did the rollback. Supposedly, this is only permitted
within 30, or maybe 10, days after the upgrade, and it's only possible if you haven't
used Disk Cleanup to wipe out the saved copy of the previous operating system.
But here's what I did:
In Settings, type "Restore" in the search box, and choose "Go back to a previous build."
Set it going. Several hours of frantic disk activity ensued.
Five hours later, the computer was sitting there powered on but with a black screen,
and the disk activity had stopped except for a brief blink every few minutes.
I decided to power it on and off (holding the power button down for 5 seconds and then
pressing it again more briefly).
The computer booted up just fine into Windows 10 build 1511 (the November Update) with
all my files intact, including newly created ones.
It wanted to download Windows updates. I let it.
I also ran the Intel Driver Update Utility and got new Bluetooth and video drivers.
Now I have a fully updated Windows 10 build 1511 system (not build 1607), and the
cameras and touchpad work fine.
I still think it's going to get a Linux partition, though.
I was dismayed at the lack of astronomy camera software for Linux, but there seems to be
just enough. And another function of this computer is to be my travel laptop,
and on unfamiliar networks it's handy to run an OS that is rarely targeted by viruses.
The symptom is that the software finds the camera and may capture video briefly, but loses the
camera connection after a few seconds or when a setting is changed.
I'm not sure I believe the technical explanation given in the first of those links because
I was not using compressed video in the first place.
I'm also not sure whether my problem came from Anniversary Update or from a later update
intended to fix it.
The first of those links gives a registry fix which did not help in my case,
but in case you want it, I'm making it available as a zipped .reg file.
Click here and use it at your own risk, preferably opening it with
a text editor first, so you can see what it does.
New drivers might fix the problem, but nobody was told that new drivers were going to be
required; webcam drivers since Windows Vista have changed little and are supposed to be upward
compatible.
I see two larger problems here:
A public relations problem for Microsoft unlike anything they've had in recent years.
Since Windows Vista they've taken a lot of criticism for strange user interfaces and
awkward controls, but not for unreliability. There had not been unforeseen changes
that would keep hardware or software from working. But now they've done it.
A failure of Microsoft's testing process. The Anniversary Update was supposedly in
wide use by Windows Insiders (beta- and gamma-testers) for several weeks. Did none of them
notice their webcams failing?
Because this whole situation didn't hit the media until August 19, I'm thinking it may be more
complex and may not have originated with Anniversary Update itself, but rather with one of the
subsequent updates, perhaps a hasty attempt to fix the first wave of webcam problems.
In any case, I expect a quick fix from Microsoft because they've broken their own Skype
application.
But I'm also exploring the possibility of doing video astronomy under Linux.
The selection of available software is much less, but
oaCapture seems to
work reliably.
A word to the wise: For it to see your camera, you'll need to
make some configuration changes as described
in its documentation.
No point of astronomical nomenclature is too trivial for me, so
I'm writing this brief note for people who are using a
search engine to find out what Vultus Irrisorie means.
If you suspect it's not quite Latin, you're right.
North is to the left.
In Sky and Telescope, September, 2016, p. 56, Sue French writes about
a star pattern called "Vultus Irrisorie, the Smiley Face, a binocular
asterism fashioned by Alabama amateur Roseann Johnston on July 21, 2001."
As you can see in the picture, it looks like a pair of eyes and a grin,
about 1.5° in diameter (suitable for binoculars),
located around R.A. 19:53.6, declination +47°17' (J2000.0).
The asterism is not actually a star cluster, although the right-hand three
stars in the grin do constitute a chain in space, all three about the same
distance from us, and the left-hand two stars in the grin also constitute
a pair in space at a different distance.
From Sue French I have more information about the discovery and naming.
Roseann Johnston's first note about the asterism, which she dubbed Smiley Face,
was posted in the Yahoo group ACATW
(Astronomy Clubs Around the World) on July 24, 2001, and is still readable
there.
Four days later, in the same forum,
Otto Danby II referred to it as
"Vultus Irrisorie, The Face with a Smile,"
with no further comments on the apparently Latin name,
which does not appear to have been picked up by anyone else immediately.
Eventually it was used by Sue French
in Night Sky Magazine in 2006, as well as Sky and Telescope in 2016.
Roseann Johnston died around 2003, and many of her friends
treated the asterism, fittingly, as a memorial to her.
What concerns me is that Vultus Irrisorie is not quite Latin.Vultus means "face" all right, but irrisorie is not a word,
as far as I can determine, or at least not anything that could modify vultus.
Irrisorie exists in Latin as the vocative singular and (with long e) adverbial form of the rare
medieval adjective irrisorius 'mocking,' from classical irrisor 'mocker,'
but Latin grammar does not allow irrisorie to modify vultus, and anyhow,
it's a very rare and far-fetched word.
In Spanish and Italian, irrisorio is an adjective that
means 'trifling,' from postclassical Latin irrisorius. In Italian, irrisorie is
the feminine plural of this adjective. That doesn't help because it doesn't mean
what's intended, it's not Latin, and
vultus is masculine singular.
I am guessing that Vultus Irrisorie was a typo for Vultus Irrisoris 'face of a mocker,' so
the asterism is the face of a mocker. Not a happy smile; a happily smiling face would
be vultus subridens or (if you want to get as close as possible to what we have)
vultus arrisoris.
Crucially, Otto Danby only typed Vultus Irrisorie in the forum once.
The phrase only appeared once in his message, which was quoted and automatically copied
a couple of times. Nobody else typed Vultus Irrisorie
with their own fingers for several years afterward, as far as I can determine.
If the original had been a typing error, it would have been preserved.
If you want to admit a nonclassical Latin word, it could equally well be a typo
for Vultus Irrisorius 'mocking face.'
A more remote possibility is that Vultus Irrisorie was meant to be humorous, half Latin
and half French, and the second word is a typo for Irrisoire (French for 'mocking').
In that case, the grammar is correct, if you allow the mix of languages.
I should add that my research is based on professional-quality sources
such as the Oxford Latin Dictionary, as well as a considerable amount of computerized
text searching; all extant ancient Roman literature had already been searched by
the dictionary makers.
The first thing that will occur to most of us is to try Google Translate.
Its Latin is notoriously bad, and, being a data-driven program, it usually claims to translate
anything fed into it, even if the input is not Latin and the result is a wild guess.
Google Translate renders vultus irrisorie as
'looks ironic,' as if vultus were a verb, and that can hardly be considered to
solve our problem.
There are those who favor letting the apparently misspelled name stick,
since it has been in occasional use for 15 years. After all, many star
names (e.g. Albireo) appear to have been mis-transcribed at some point in their history.
My own preference, however, is to use the name that Roseann Johnston herself used,
Smiley Face.
P.S. I know this is pedantic. If it weren't, it wouldn't be historical linguistics!
Why did I give you a picture generated with Stellarium rather than a photograph taken
with my own camera? Because we've been clouded out for weeks, with a thunderstorm every
evening, sometimes as late as midnight!
One of my linguistics colleagues recently pointed out that in some parts of Asia, non-native speakers
of English, who use it constantly in daily life, far outnumber the native speakers.
Consider India alone, with its enormous population. English is the language of wide-area
communication there among people who did not learn it at their mother's knee.
That brought to mind the historical importance of languages without native speakers.
Latin was the language of international communication in Europe for 1500 years after
there stopped being native speakers. Sanskrit had a similar role in India.
The lesson? The native speakers of a language are not the only people who use it; at times,
not even the most important people who use it.
I think I've figured out why so many people, with varied backgrounds,
enjoy doing computer support and related technical work for a living.
It overcomes the sad separation between "management" and "skilled labor"
that makes most white-collar jobs dull.
Traditionally, only blue-collar workers regularly experience the
satisfaction of making things and fixing things.
Every plumber and carpenter does, but few managers do.
Managers mostly run things (maintaining the status quo)
or make very abstract things (organizational structures,
programs, and services), any of which can be taken away by
the stroke of a higher manager's pen.
Computer geeks get to make things and fix things that other
people can use and enjoy, without losing
white-collar prestige.
For five minutes today (Aug. 19), at the beginning of a power failure, the
AC line voltage in our house dropped to 60 volts and stayed steady at
that level.
It was interesting to observe the symptoms. Incandescent bulbs and dimmable
LED lights continued to work at reduced brightness.
Some flourescents flickered on and off at about 1 Hz (not corresponding to
voltage fluctuations — the voltage was steady).
Low-power devices powered by wall warts continued to work normally,
including my laptop.
Larger pieces of computer hardware shut themselves off.
And our HE washing machine stopped, then intelligently resumed when the
power came back on.
I'm not sure how we came to have exactly half the normal line voltage.
Possibly something involving half of a transformer winding burning out —
I'm not sure. I know they got the power back on after just 20 minutes
and texted me to say so.
Sharon's computer fell victim to an unusual Windows 10 bug that I have also
heard of in Windows 8. The keyboard became completely unresponsive. The computer
still responded to the mouse (and touchpad) in the normal way.
The cure? Hold down the Windows key and press the space bar.
If you have more than one keyboard layout installed, Win + space is how you choose among
them. For example, I have U.S.-English, U.S.-International
(very handy with foreign languages), and two foreign alphabets installed on mine.
But if you have only one, apparently the other option becomes "no keyboard at all"
and the only thing that will wake the keyboard up is Win + space.
What I was calling the carnage on
the third floor — that is, the announced elimination of most of the library stacks that
most interested me — turned out to be largely a false alarm.
Apparently, announcements of books to be removed temporarily for construction were misunderstood
as announcements of books to be removed permanently.
About 80% of the books still seem to be there, including many as far back as the 1940s.
What's missing, right now, is Q (general science), QA1-QA10 (very general mathematics),
G (geography), and some of the latter letters under T (an assortment of odd areas of
engineering). But there are also enough empty shelves to make me think a good bit of
material is yet to come back.
And (to make a very erudite joke) the transformation matrix has an eigenvector:
there is a shelf of books around QB1 (astronomy) that seem not to have moved at all!
Of course, it has been weeded out. QST from the 1920s is no longer on open shelves.
Although I have historical and nostalgic tastes, I willingly admit that a good bit of older
material is not acutely needed today.
What I worry about — at all libraries, not just here — is what I call the
"digital dark ages," scholarly journals from about 1980 to 1995. Those were too old to get
digitized, but they had too short a life, before the digital revolution, to get cited and
become known by indirect channels. They are doomed to be read less than journals either
before or after.
Getting back to the library: Large study areas were created, but I'm not sure the capacity
for books was actually reduced; the book stacks seem to be using the space more efficiently.
The study areas look very comfortable and include abundant sources of electric power.
That's what was wrong with the second-floor remodeling a few years ago: they overestimated
the need for University-supplied PCs (not widely used at all) and underestimated the need
to charge laptops.
As I said three months ago: To this familiar space, vale atque ave.
With older video hardware, use Linux Mint MATE, not Cinnamon
On Minerva, the older computer that I use as an "operating system zoo,"
the main operating system is Linux, but it can also boot Windows XP and
DOS 5.0, and I'm leaving room for Windows 3.1 and OS/2 Warp.
I made the mistake of updating Ubuntu Linux from a version that had some
trouble with my old video hardware (Intel on the motherboard) to a version
that wouldn't work.
After a few false starts, I found out what I needed:
Linux Mint MATE, a derivative of Ubuntu,
which is a derivative of Debian.
Unlike the more popular Cinnamon interface, and Ubuntu's various new interfaces,
MATE does not require hardware video acceleration.
Apart from that, it looks very much like Cinnamon, which looks
a lot like Windows 2000 and Vista. It's convenient and familiar.
Here mate is a 2-syllable word,
the Spanish name of a South American tea-like beverage.
I have no idea whether people put mint in it.
I know that Mint Cinnamon sounds like a definite clash of flavors.
Two technical tidbits about the new Ubuntu Linux subsystem in Windows 10...
(1) There are two filesystems. The main Linux filesystem is stored in your
local APPDATA folder and follows Linux (UNIX) rules completely.
Under /mnt you see your Windows drives, and Linux uses a slightly
different filesystem that requires filenames to be Windows-legal.
There, you are free to work on the same files with Windows and Linux (in fact,
that's the point). Linux treats the filenames as case-sensitive; Windows,
case-preserving but case-insensitive.
If you create TEST.txt and test.txt in the same Windows directory,
Linux will see both of them, but Windows will have trouble distinguishing them
(it may or may not, depending on the exact context, and seems to favor the one
that was created first). You have been warned.
(2) To achieve UNIX-like performance on fork and similar operations,
Windows has introduced a new kind of process, a "pico process," which has much
less OS overhead than a traditional Windows process. To Windows, a pico process looks like
an allocated block of memory with nothing going on in it.
That's the question we were really wondering about, and now you know the answer.
Just before Emily was born, I got two other pieces of news that should have been noted
here but got pushed out of the way.
First, happy retirement to Don Potter, who was director of the Institute for
Artificial Intelligence during my last few years there. We have now had a complete
changing of the guard — but of course the Institute is still run by people we
know and trust. We wish Don a happy future.
Second, there will be much less carnage on the Third Floor than I was previously told.
That is, the third floor of the UGA Science Library will not lose all the books
dated before 2000. About 75% of the books will stay, though some were temporarily removed
during the remodeling.
I saw the remodeled space the other day and will surely use it a lot. It's similar to the
modern part of the first floor, but with a lot more electrical outlets for laptops.
What if there is no such thing as a rule of grammar?
I owe much of my fame to the Swedish linguist Joakim Nivre,
whose book
on dependency parsing gave abundant credit
to some of my earlier work.
He even came out with a "Covington parser," which is named after me.
What in the world, you ask, is dependency parsing?
Parsing is the act of finding out the structure of a sentence.
Computers have to do this in order to understand human language.
Humans do it too, when they're teaching or studying languages.
Presumably, we all do it automatically in our heads whenever we
comprehend or produce a sentence.
Dependency parsing is a particular kind of parsing (shown above) in which
the structure is described as word-to-word links.
In my (in)famous 2001 paper, I ignored the role of grammar — I just assumed
that a computer, parsing a human language, would have some way to find out which
word-to-word links are permitted. I wrote about how to organize the overall task
of making all the links.
There Joakim Nivre stepped in, to do the part I hadn't done. And that's where
it gets interesting.
Unlike most parsers, his doesn't have rules of grammar.
Instead, it has a knowledge base of what words it has seen in what contexts
and what kind of links they had.
It was "trained" on a large collection of English sentences that had been
marked up to indicate the sentence structure.
And it works. He only achieved about 85% accuracy with an early version, but then,
it was trained on a fairly small (multi-million-word) collection of sentences, and
they were not necessarily all marked up correctly.
A human being's knowledge of his native language is based on much more data,
with more opportunity to find out what sentences mean and reject the ones that may
have been misunderstood.
To make decisions while parsing a sentence, Nivre's parser doesn't look at rules of
grammar. It looks at the knowledge base and asks itself, "What have I seen that
resembles this, and what kinds of links did it have?"
Old-fashioned historical linguists used to tell us that language works by analogy,
and as far as Nivre's parser is concerned, that's absolutely right.
What if the parsers in our brains work the same way? What if (contrary to what
Chomsky and numerous linguists in his tradition have always thought) there are no
rules of grammar in our heads (except a few that we were taught in school)?
What if our brains actually construct sentences (for comprehension or production)
by asking themselves, "What have I seen that is similar to this?"
What if categories like "noun" and "verb" are simply shallow generalizations that
our brains make in order to treat large classes of uncommon words alike?
"I haven't seen this word very many times, but it looks like it's used the
same way as these other words..."
Notoriously, the grammar of common words is messy. Rare words are well-behaved,
but common words develop quirks all their own.
That is exactly what you would expect if our brains work like Nivre's parser.
If we see a word often, we have plenty of examples of it and can use it in an
ever-widening
variety of ways. If a word is rare, it has to be lumped with other similar words.
When I say there may be no rules of grammar,
I don't mean nothing is ungrammatical or that it's OK to string words
together randomly.
Sentences definitely do have definite structures. (They could hardly have
definite meanings without it.)
Languages allow particular structures and not others.
Above all, there is definitely such a thing as vocabulary —
and it does more work than we gave it credit for — it stores our experience and
knowledge of every word in the language.
I am introducing this only as a possibility to be explored.
Clearly, there are linguistic phenomena that are hard to explain without
grammar rules, such as WH-movement.
But let's keep data-driven sentence structure in mind as a possibility.
Today was a very big day of another sort (which is why we're not with Emily).
Specifically, today was D-Day in the battle against the chronic nerve pain that
my daughter Sharon has suffered for five years.
We don't know what happened, but somehow, in the spring of 2011, Sharon suffered
damage to a thoracic nerve, producing chronic pain of disabling severity.
The exact nerve was identified a while back, and the damage was probably done by
a viral infection. Nonetheless, we've had a hard time dealing with people who were
sure Sharon's pain was psychological or even fake. Word to the wise:
There are people in the world who are in severe pain even though
you can't see the injury.
The pain was very hard to relieve; it is too peripheral for opioids and too
central for Motrin. So she sat and suffered.
Today we went to a
surgeon in Covington, Georgia
(the town where everybody knows how to spell our name)
who had been recommended by Sharon's local spine specialist,
and a Boston Scientific spinal cord stimulator was implanted.
This had already been done on a trial basis in May and worked very well.
Today, the operation was a success, but she will not get the full
benefit until she has healed and the programming of the device
has been adjusted.
We hope this is the beginning of a new life for Sharon.
Windows icons should be 32-bit, not 24-bit, if they involve transparency
Windows icons that involve transparency (i.e., that have a transparent background
rather than filling every pixel of the square) should be 32-bit rather than 24-bit
(a.k.a. 8-bit). That is, they should include an alpha channel (transparency value
for each byte).
Otherwise, diagonal and curved lines bordering on transparent areas
will have a stairstep appearance. Windows can't
antialiasbetween the icon and the desktop background. However, your own graphics
creation software can do so, and should, by using an alpha channel, so that a given pixel
can be, for instance "40% this color and 60% whatever is under it."
My preferred workflow is to create the icon (256×256 pixels) in Photoshop
(File, New, background: Transparent) and save it as a .PNG file, which preserves the
transparent background nicely. Then I open it in IcoFX and convert it into an icon.
Automatically, IcoFX creates icons in all sizes based on the initial 256-pixel image,
but you should delete most of them. Keep 256, 48, 24, and 16-pixel icons, all 32-bit
(not 24-bit, also known as 8-bit, or anything less). Then perform any sharpening or
editing that you think appropriate.
Windows 10 uses the 256-pixel icon as the source of other sizes needed.
It displays 48-pixel icons on the screen, 24-pixel icons on the taskbar,
and 16-pixel icons in the corners of windows.
I'm getting pretty good with spackle and sheetrock mud. This is where the
upstairs thermostat used to be. It took more than one pass, but using spackle,
sheetrock mud (drywall joint compound), and paint, I was able to get rid
of a one-and-a-half-inch hole in the wall.
It helped that, years ago, we chose to paint the walls pure white
("Scandinavian white" as some decorators call it).
That makes it easy to match. Behr Ultra Pure White ceiling paint
happens to be perfect for touching up the walls.
For years, I tried to do this kind of work exclusively with spackling.
Results were never perfect because spackle dries harder than sheetrock
and requires sanding. When I would sand it, irregularities would appear,
due to differences in hardness. The trick is to do the final coat with
sheetrock mud, which is softer, can be spread perfectly smooth with a wide
putty knife, and does not require sanding.
For a long time, Windows has been POSIX-compliant in various ways, but now, with yesterday's
big upgrade, it has a full Ubuntu Linux runtime system.
Designed for developers who have to develop Linux software on Windows machines,
the new system will also be very useful to those of us who use UNIX-based scientific software.
Here are the main things to know:
It's real Ubuntu Linux, from Canonical, the maker of Ubuntu, which puts it in the Debian family.
It downloads packages from Ubuntu using sudo apt-get.
To get it, you have to turn on developer mode (to be able to install non-certified apps
from the App Store) and then select it as an optional Windows feature.
More here.
It's command-line only.
I figure the Linux community will very quickly find a way to support graphical
Linux applications, but that's not there yet.
It uses the bash command shell, hence its name. You can of course install tcsh.
It includes the pico full-screen editor, for those of us who don't like vi.
You can easily install Emacs too.
This is not a set of UNIX commands for a Windows command prompt.
Those are UNXUTILS and I still think you need them.
This is not a virtual machine or "sandbox." It is a full Linux API using the Windows kernel.
This is not kernel-mode Linux, so it can't use Linux device drivers.
This is also not an emulation layer such as Cygwin. It's real Linux under real Windows.
It runs genuine Linux binaries. You can use gcc and similar compilers.
It can easily access all your Windows files, as in the example above. It's easy to use
Linux software to work on the same files that you work on in Windows.
The Linux installation is per-user; it resides in your local application data folder.
Every user of your Windows machine can have a separate Linux with different software installed.
This may be the future of Linux — as a subsystem under other operating systems, making highly
portable software possible. Linux is also mixed with the operating system of IBM mainframes
(exactly how, I'm not sure; and an IBM mainframe is now a box the size of a PC).
Afterthought: What does UNIX/Linux buy us? Cumulative development.
Linux and BSD today can still run, unmodified, scripts and programs that were written in the early 1980s.
For those of us with long-term scientific research projects, that can be very important.
And it also gives us portability; the same programs can run on a wide variety of computers.
The Macintosh has been UNIX-based for some time. Now that Windows has an (almost) full-fledged
Linux subsystem, the door is open to even wider software portability. If only it had a GUI... and
that is probably not far off. I hear of successful experiments with the X Window System already.
If you have just gotten the major Windows 10 update ("Anniversary Update"),
I strongly recommend checking the
security settings that I wrote about last year.
Microsoft continues to do some things with networking, by default,
that I think are bad decisions.
If you go through the "Welcome to Windows 10" process on your first login,
DO NOT choose "Express settings." Instead, choose "Customize" and answer "no"
to nearly every question (it will be obvious what each of them is about).
In my limited experience, if your first login does not
going through the "Welcome to Windows 10" process, your settings will remain
unchanged from before. Even so, it is a good idea to check them.
As best I can determine, you get "Welcome to Windows 10" (and reset settings) if
you go to Microsoft's web site to download the Windows 10 installer.
If you get the update automatically, you bypass that step.
Here's another series of pictures taken with the AT65EDQ (6.5-cm f/6.5) refractor
that I'm trying out. Above you see a stack of ten 1-minute exposures of M17
taken with the telescope on my AVX mount and unmodified Nikon D5300 camera.
PEC was used but no guiding corrections were made. I took advantage of a hot
but unexpectedly dry and clear night. The Milky Way was visible from my home —
which it had not been for many years — thanks to our town's recent conversion
to LED streetlights that don't shine sideways or upward.
This is the central part of the image, originally 2400 by 2400 pixels, reduced by
a factor of 3 for display here.
I know I've photographed this object many times before, but this is, after all,
an equipment tryout. I actually took this series of pictures the night before
the ones I published a couple of days ago.
"Lightning flashes" or other graphics problems in Stellarium
I use, and strongly recommend, Stellarium sky
mapping software, which is free. The images of the sky can be set so that they are realistic
enough to use in planetarium projectors.
A new version, 0.15.0, has just come out, and when I installed it on my older Lenovo laptop,
I got a thunderstorm! That is, there were often lightning bolts flashing across the screen.
(Definitely not meteors, which Stellarium can simulate.)
This was actually a graphics problem relating, apparently, to pixels just outside the displayed
image. Zooming changed it and would even make it go away for a while.
The cure? Nowadays Stellarium gives you several startup links that use the
graphics hardware in different modes. I had to use Stellarium ANGLE WARP
instead of Stellarium (and pin it to Start). I don't know what ANGLE WARP means,
but it works.
This is a private web page,
not hosted or sponsored by the University of Georgia.
Copyright 2016 Michael A. Covington.
Caching by search engines is permitted.
To go to the latest entry every day, bookmark http://www.covingtoninnovations.com/michael/blog/Default.asp
and if you get the previous month, tell your browser to refresh.
Portrait at top of page by Krystina Francis.
Entries are most often uploaded around 0000 UT on the date given, which is the previous
evening in the United States. When I'm busy, entries are generally shorter and are
uploaded as much as a whole day in advance.
Minor corrections are often uploaded the following day. If you see a minor error,
please look again a day later to see if it has been corrected.
In compliance with U.S. FTC guidelines,
I am glad to point out that unless explicitly
indicated, I do not receive substantial payments, free merchandise, or other remuneration
for reviewing or mentioning products on this web site.
Any remuneration valued at more than about $10 will always be mentioned here,
and in any case my writing about products and dealers is always truthful.
Reviewed
products are usually things I purchased for my own use, or occasionally items
lent to me briefly by manufacturers and described as such.
I am an Amazon Associate, and almost all of my links to Amazon.com pay me a commission
if you make a purchase. This of course does not determine which items I recommend, since
I can get a commission on anything they sell.
If what you are looking for is not here, please look at
previous months.
