ever hear of killing explorer from task manager and restarting it from task manager?
Linux still has the edge over that I believe. AFAIK, killing Explorer will not kill "child processes" since they aren't really child processes and re-starting Explorer will not re-run some of the startup processes since those are a side-effect of a login, not of Explorer starting. As a result it is common to lose notification icons while the processes that owned them continue to run. MS provided a protocol for recovery from this (after it became obvious it was necessary I believe) but I've seen few examples that use it. It is probably more accurate to add a logout and login to the description of the recovery process since that will usually recover the proper quiescent process state. I have to concede that I side with the opinion that it is more likely that you will be able to recover from a UI lockup without a reboot on Linux than you would on Windows.
NB: The following implies nothing in relation to your comments and is intended to reflect only my opinion of the whole set of replies to the story. FTR, I mostly use Windows, because I mostly like it. My employer deploys a Windows desktop internally, I have work that requires Windows and Linux workstations and use both daily, though the Linux hosts are nothing like the "policy Linux desktop". My impression, based on my experience, is that the Linux rebuttal to the Windows assertions in the replies to this story is largely accurate and balanced whereas the Windows assertions seem, well, paranoid, very defensive and somewhat desperate in tone.
Canon also limits what you can do with their Raw and Digital Camera SDKs. I've had an EOS-10D for 13 months and shot about 2500 frames. I shoot RAW and I needed some software to convert to useful formats for web publishing and in multiple sizes. I found nothing so started my own shameless plug. I got the Canon SDK only to discover it doesn't do half of what I needed. For example, with the SDK you can't get camera body rotation from RAW files as far as I can tell. I had to research the file format and work it out the hard way. You can't convert images to other formats in memory. I read one RAW and generate JPEGs in three different sizes (sensor, suitable for dial-up retrieval, and thumnail). The Canon SDK forces me to convert all images to TIFF, load the TIFF once and convert it in memory three times. BTW, the EOS10D claims to provide discrete shutter speeds and apertures but the RAW files record intermediate values. Not surprising that the camera has finer TV and AV resolution that the camera UI suggests really but annoying to handle in imaging software.
Hear, hear. On paper Star Trek is the story of a communist future. Marx proposed a utopia like that in Star Trek as I understand it. Communism was turned into a dirty word by dictators, not by communists.
...at least from Microsoft perhaps. This sounds like a straight copy of Novell's data protection server feature and, no doubt, a few other technologies from various platforms. I wonder if they'll even copy the Explorer shell integration? I think the Novell one is a bit more than RSync and traditional backup but is limited to Novell's file system (NSS) as I understand it. It's more like CVS meets backup. It is disk to disk but the target disk isn't a duplicate of the source (i.e. not COPY, XCOPY or RSync). The target is a database of versions. I think Microsoft's is similar. In NSS the file system internally tracks changes and can give you them in a list (doesn't need to be trawled for changes). It supports snapshot with copy on write at the file and volume level so you can exclusively lock a file without preventing it from being changed (though you have to have some knowledge of concurrency among your applications to use it safely). Files retain all security attributes in the archive. Users can do their own restores from their workstations (Windows ones get a context menu option) and their access rights are enforced in the retore software so they only see files for restore that they had righs to originally. The common restore case: single file due to user error I believe. If users can restore their own files you don't need to recover a tape, cue it and wait for it to spool to the file in question.
Someone did succeed in defending themselves against criminal charges relating to some use of their PC based on trojan horse software found on their PC. The defence was, IIRC, that the trojan was responsible for the criminal activity. That's what made me think this was an effective get out of jail free card.
My understanding is that civil suits have a lower burden of proof and I believe the RIAA is using the civil process. But my guess is that it would be an argument with a fair chance of success, especially before a jury of average PC users. It would probably come down to whether or not the RIAA could do a hands on analysis of your PCs. If so then they'd find the MP3 files and the playlists that refer to them. You could try to argue away the presence of the files as the hacker's use of your storage but you'd have trouble with the playlists I guess. I have a lawyer friend that I might ask about this.
The OP said it wrong I believe by saying "sin(x) can have values between positive 1 and negative 1" and failing to add inclusive. 1 isn't between 1 and -1 and -1 isn't between 1 and -1.
At least since the OP said his/her x was a value between zero and pi and didn't add inclusive then sin(x) is positive, it might have been zero (is that positive or negative) if the OP had said x was between zero and pi inclusive.
</excessively pedantic>
So, just about any law you can break with a computer is now fair game. When you go to court just refer to the three minutes it could have taken some nefarious hacker to use your network without your knowledge. Since the likelihood of such an attack is low then I recommend everyone use a dictionary entry to generate keys. It will keep your neighbours off your network and you'll leave yourself with a perfect reasonable doubt defence when sued or prosecuted.
Is it just me or is the Groom Lake imagery lower resolution than other parts of the country. Also, you can't zoom in as far as you can in other parts of the country. For the spooks watching, I merely tried this in order to be able to report any breach of national security to the appropriate authorities. Now where's that tinfoil hat...
NPR (and NPR "member" stations presumably) and the BBC World service are "members" of PRI as I understand it. Members having the same interpretation as in NPR "member" stations. I frequently hear the phrase "PRI member" referring to individual NPR "member" stations and NPR (particularly the NPR news organisation) used during "The World", a daily PRI show broadcast by NPR "member" stations.
Add to that the BBC (along with NPR, a member of PRI). All of the BBC radio stations have content available for 'Net re-broadcast (I believe you have to record them while playing as well). Radio 4 has excellent speech content with some fine comedy alongside in-depth art, science, current affairs and analysis (e.g. political interviews with members of both sides of an issue in the same studio at the same time).
Some of the BBC music stations are pretty good too. Many of them are segmented by market the way that US radio is but none of them have the sort of motivations that make much of the US radio I have heard just crap (IMO). Long live NPR!
If the other thing from the other company is worth what you pay for it then you pay nothing for the iPod. Arguably buying the other thing from the other company is "authentication" only for the free iPod. You did want that BowFlex, matching luggage and super-sized viagra shaped Rolex watch didn't you?
Re:If you want a time-lapse capture app for free..
on
Infrared Webcam HOWTO
·
· Score: 1
I'm scared to release the source, though I have thought about it a lot. It's mostly hubris that stops me, plus I'd like to avoid the inevitable abuse about my style and suggestions about how to implement it "properly";-) It's hard knowing which kudos to chase: the prospective employer with a million dollar offer for someone that can do this work; or the community. I'm a utopian with a family to feed. As an "anonymous coward" poster you might understand me being a named coward.
The DirectX SDK has some very good Direct Show examples and I used them for the source to real-time display parts of WebCam. There's a little filter builder test tool in the DirectX SDK that lets you chain elements to see what goes together to connect a source to a window.
WMV was an absolute bitch. The examples in the SDK are just rubbish and I ended up doing bizarre things like writing a program to dump the entire state for all of the installed Windows Media profiles in order to find out some of the undocumented properties that are effectively necessary for decent results. The default profiles are fine if you are using "Windows Movie Maker" but not for reasonably configurable applications.
GDI+, OTOH, is just fabulous (to program with). It's pretty well documented and examples aren't really necessary due to its simplicity. Though it's not without its own stupid behaviours, e.g. inability to construct an object to create an EXIF tag (though I haven't looked into whether or not that's just a feature of the.NET framework GDI+ interfaces). One of the other apps on my downloads, Thumber, heavily uses GDI+. That introduced me to another SDK that is just rubbish: the Canon SDK for the RAW image format from its digital SLRs. It's functional but gives very poor access to files.
Anyway, I'd prefer not to publish the source but you can always mail me and I'll discuss the elements of it and address any questions. Send mail to the dmair mailbox in the mair-family dot org domain.
Er, wasn't the point that the useful (intended) doppler shift comes at the cost of another doppler shift in the signal coming from the camera to the viewer. IOW, when there is blue shift that moves the IR into the visible spectrum as the camera is moving towards the subject the camera is flying away from the receiver so fast that there is a red-shift in the signal from it. That was my understanding at least. FWIW, I didn't know if that would affect a cabled webcam or only a wireless one (obviously for a cabled camera you'd run the wire through a collar at the hob of the rotating parts).
If you want a time-lapse capture app for free...
on
Infrared Webcam HOWTO
·
· Score: 4, Interesting
...Get one here: http://www.mair-family.org/Downloads/ Geoff says the software he had isn't very good. I couldn't find anything I was prepared to pay for so I wrote my own and publish it at that link. Sorry Windows only and WMV output. You can read the feature list for yourself but, in terms of Geoff's difficulty, it will capture frames at between 4s and one hour per-frame (configurable) and create a 15fps WMV.
Go ahead,/. my web server...Hopefully my ISP is tolerant.
Despite the growth in broadband capacity I am unconvinced that it will have the capacity to support HD video at DVD like quality for multiple homes in a single neighbourhood any time soon. Rotating HD capable media will arrive a long time before that and the players will play your existing DVD collection. Sounds like a hard habit to break.
In places like the UK where the wiring infrastructure is ancient it's hard to see how broadband will ever develop sufficiently for complete replacement of physical media (despite the BBC and UK government's plans). Wireless may resolve that but not any time soon and not at all for many homes. This is true for many European, American and African countries and many cities in the USA, all of which can support some form of DVD market. For the studios this means supporting more than one distribution medium versus DVD/future DVD where one disk fits all (region code notwithstanding).
A SMTP proxy may be as good: http://sourceforge.net/projects/assp/ I use this configured to give an error response on delivery of messages it rejects. It receives the whole message first and all spam goes into a dump mailbox. So I get to monitor it's filtering and cause failures on incoming spam. FWIW, I've noticed my daily spam message counts decreasing since I enabled error on spam behaviour.
Segments exist in 32 bit x86 flat memory model environments too. Code is loaded from cs:eip and data is accessed at ds:offset (there are several modes for determining offset). A flat model makes it possible for all references to be near (no explicit segment) but that only means cs is implied (code) or ds is implied (data) es may be implied too for string instructions. The OS loads cs once and ds(,es,fs,gs) once for any user mode process but segments are still loaded and used.
NB: I'm talking about what you can do (in response to you your "could not" point), not what you might prefer to do when implementing an OS (I don't like segmentation either).
I don't believe it is true that you can't make memory readable and not executable but AFAIK these exploits depend on writing and executing anyway. I believe you can make addresses executable but not writable on an x86 without NX. Segmentation allows you specify a base address and limit for a descriptor that can be selected by cs (what eip must be an offset in). If you have no writable data descriptors that overlap any code descriptors then a user mode application can't execute arbitrary code by poking instructions into memory. IIRC, you'd get a GPF trying to write opcodes where you could execute them or a GPF attempting to jump to an address where you could write opcodes.
You can also make code readable by putting it in read only pages but, if cs has a 4GB limit, you can't prevent a program from running arbitrary code it pokes into memory it can write to. I think that is the real cause of the problem and the OS chooses the values of the base and limit for cs. NX provides executable control in the paging system but it has been available in the segmentation system since at least the 386 (can't remember 286 descriptors). You have to use segmentation even in a flat model (implicitly) but you don't have to use paging. That (and MS bugs) may be a factor in the time it has taken to develop NX.
FWIW, I don't advocate segmentation and an executable page attribute has been missing from the x86 MMU since the 386. IMO it is an omission in the x86 design that you can have supervisor and user segments in read, read/write and executable types, but supervisor and user pages in only read and read/write types.
I don't actually know how they measured it (and I'm not an astrophysicist and don't offer this as opinion, let alone fact) but 1.5 million miles per hour probably creates a significant doppler shift. The spectra of stars shows distinct bands that correspond with the absorption of photons at specific wavelengths. The reason the absorption is at specific, discrete wavelengths is that electrons in specific elements (e.g. Hydrogen, Helium, etc) absorb at specific wavelengths. The distance between absorption lines for different elements is fixed. For example, say Helium absorbs at some wavelength x and Hydrogen at some wavelength y. The difference between x and y is a constant (even under a doppler shift). So, if you see two absorption lines and they are x - y apart then one is the Hydrogen and one the Helium. This is overly simplified but the principle is the same (it just involves more than two lines). You can then tell how much doppler shift because you know what wavelength they should appear at and see what wavelength they actually appear at. Doppler shift is directly proportional to speed (I hope that's accurate). So, knowing the doppler shift you know the speed relative to the point of observation. Presumably you can work out the speed of the earth relative to some object by measuring the relationship between doppler shift from that body and the doppler shift from various other celestial bodies.
Or, maybe they had a very long tape measure, a very powerful radar, or a very powerful laser or just guessed;-)
The Boeing computers auto-tune multiple VOR receivers (actually uses DME as I understand it) to correct INS drift. If you are receiving from two VORs then you can plot an arc of possible positions, three and you can plot a point in space. Then it's only a matter of accuracy in the system. If you look at the Boeing ND with stations displayed then the auto-tuned VORs are highlighted in blue.
Spoke to my CFII/ATP friend. WAAS (augmented GPS) approaches are appearing regularly. While they are precision approaches none of them are zero visibility qualified. They are about 10m vertical less accurate than equivalent ILS approaches (that's a 20 foot minimum threshold crossing versus 50 for an ILS) and there is a consequential limitation on glideslope angle that makes zero visibility landings impractical (steeper approach, flare up to 30' too high or too low). The ones we looked at were all slightly worse minimums than an equivalent ILS approach (not Cat III), for the reason above presumably. I can only remember the RVR was 1.5 miles but I can't remember the DH, I think it was 500'. I guess this doesn't prevent them from being used for auto-land however (you can always hand fly once you can see the runway). My friend tells me that the FAA is actively giving vibes that the status of WAAS GPS approaches are not going to be improved upon (period). I get the impression that the FAA doesn't trust GPS unless you can see the things you might hit.
Radio altimeters, yes. The 747-400 will talk to you. If you want it to it will call out the reference speeds on take-off and useful altitudes on approach. It also tells you when it isn't happy with your flying! Listening to a 747 barking "GLIDE-SLOPE, GLIDE-SLOPE" then a minute later "TERRAIN, TERRAIN, PULL-UP" because it is unhappy with the radio altimeter reading makes for really sweaty palms.
Aircraft with engine out (including gliders) always have right of way (I think that non-steerable aircraft, e.g. balloons, maybe have a higher right of way). I once had to call another aircraft on approach to get them out of my way when I was engine out and my instructor had to go on the radio to insist that they comply with the regulation they obviously didn't know. Helicopters look like fun but are really expensive.
Just out of curiosity, is the Tacoma approach in question WAAS-only, or just IFR-GPS?
It's an old chart so it is VOR-GPS only.
I also don't follow the logic about the reference point being between the two runways. If the computer knows where it is, and where it's going, any 'reference point' (which is actually provided by at least 4 sats possibly plus a ground station) is completely arbitrary. And if the autopilot can't (physically) fly it GPS, it can't fly it ILS, either, right?
Yes and no I guess. The published procedure is what the crew/aircraft is supposed to follow. For the VOR-GPS plate I looked at the GPS approach is a GPS overlay of the VOR approach. The approach is a runway heading to the VOR but the VOR is not on either runway, it's between the two (nearer the right from what I can tell). So, the published procedure is to fly down to that VOR on the runway heading but to the right of the left runway or to the left of the right runway. Then, when the field is in sight, side step to the correct centreline and perform a visual landing. That appears to make it largely a locating procedure because you may as well line up on the correct runway as soon as you are able (gets you stabilised as early as possible). There's no guided profile (glideslope) so the approach is officially non-precision.
I once had the good fortune to get a few hours on a 747-400 Level D simulator (a BA 747 training captain needed help with his laptop). I got to watch two first officers doing a dry run of the check-ride they were doing the next day. It was a wild set of coincidences they were given but I got to see them both do a non-precision NDB approach into Gatwick. The ceiling was 600 feet AGL. The pilot flying was calling the distance to run every mile using the navigation display and the pilot not-flying would call the altitude for an on profile approach. It was obvious to me from the ND that the aircraft could have handled the approach better using the INS and VOR augmentation than either pilot (particularly one of them) was able to by hand flying using the ND but regulation and policy prohibited it. The aircraft is only pre-programmed with the published procedures (plus it can probably handle straight in using the ILS without needing a procedure - select the ILS frequency, select the auto-pilot in approach mode and sit back). So, if you want to use the auto-pilot for an approach and landing your only options (for the types I've had any kind of access to) are to hand program the approach waypoints (too slow and prone to error plus, at least the 747-400, won't auto-flare) or to select a published procedure from the flight computer. If the only published GPS procedure is a non-precision approach (no glideslope) or has no guidance onto the runway (the VOR-GPS at Tacoma for example) then regulations (and probably company policy and auto-pilot implementation) will prevent you from using it as a zero visibility approach. FWIW, company policies often enforce stricter limits than those enforced by the FAA. Incidentally, cool 747-400 software bug... The BA pilot that let me fly the sim told me he once did a re-positioning flight to move a 744 from Gatwick to Heathrow (less than 10 minutes flying time). He tried to tune the Heathrow ILS frequency on the radio only to discover you couldn't select an ILS within 10 minutes of a take-off. He assured me it's been fixed.
I know that typical GPS approaches have worse minimums than comparable ILSs, as the (unaugmented) accuracy is not as good, particularly when it comes to altitude. I'd have to check, but IIRC, WAAS is supposed to be accurate to within 1.5 meters. And believe me, I (and many other students) have dropped a plane out of the sky from higher than that.:)
I did glider training once. Flying a three degree glideslope is too risky so you do ten degrees or something equally wild. It's about the best roller coaster I've ever ridden. My instructor (the friend I've been paraphrasing) used to shout at me f
I'm currently looking at the Cat III ILS RWY 16R approach chart for Seattle Tacoma. The Straight in ILS 16R (S-ILS 16R) minimums for CAT IIIc (any aircraft category) is N/A (i.e. no runway visual range limit - i.e. zero visibility). The Cat IIIb has a RVR limit of 300 feet. None of the listed Cat III minimums includes a decision height. The following is written on the chart:
"CATEGORY III SPECIAL AUTOLAND EVALUATION REQUIRED:
"CATEGORY III ILS-SPECIAL AIRCREW & AIRCRAFT CERTIFICATION REQUIRED"
My understanding is that a Cat IIIc approach and landing is a coupled approach only and approach and landing are permitted in zero visibility.
I also checked some Cat III charts for Denver and Salt Lake City. All have no RVR limit or decision height for a Cat IIIc approach. All have the same certification notice. Most of the runways in question have multiple ILS approach charts, several non-cat III plates and one Cat III plate.
The GPS approach I looked at, Tacoma GPS/VOR 34L/R for example, has a lowest decision height of 840 feet and RVR of 240 feet. The Tacoma GPS approach I looked is for both 34L and 34R so it has to require a visual final and landing since the reference point for the final approach course is between the two runways. Even with better accuracy than is currently available from GPS an autopilot couldn't fly that to landing.
Slashdot Mirror
Further research showed intellegence was higher in those that typed slashdot.org into their browsers.
Linux still has the edge over that I believe. AFAIK, killing Explorer will not kill "child processes" since they aren't really child processes and re-starting Explorer will not re-run some of the startup processes since those are a side-effect of a login, not of Explorer starting. As a result it is common to lose notification icons while the processes that owned them continue to run. MS provided a protocol for recovery from this (after it became obvious it was necessary I believe) but I've seen few examples that use it. It is probably more accurate to add a logout and login to the description of the recovery process since that will usually recover the proper quiescent process state. I have to concede that I side with the opinion that it is more likely that you will be able to recover from a UI lockup without a reboot on Linux than you would on Windows.
NB: The following implies nothing in relation to your comments and is intended to reflect only my opinion of the whole set of replies to the story. FTR, I mostly use Windows, because I mostly like it. My employer deploys a Windows desktop internally, I have work that requires Windows and Linux workstations and use both daily, though the Linux hosts are nothing like the "policy Linux desktop". My impression, based on my experience, is that the Linux rebuttal to the Windows assertions in the replies to this story is largely accurate and balanced whereas the Windows assertions seem, well, paranoid, very defensive and somewhat desperate in tone.
Canon also limits what you can do with their Raw and Digital Camera SDKs. I've had an EOS-10D for 13 months and shot about 2500 frames. I shoot RAW and I needed some software to convert to useful formats for web publishing and in multiple sizes. I found nothing so started my own shameless plug. I got the Canon SDK only to discover it doesn't do half of what I needed. For example, with the SDK you can't get camera body rotation from RAW files as far as I can tell. I had to research the file format and work it out the hard way. You can't convert images to other formats in memory. I read one RAW and generate JPEGs in three different sizes (sensor, suitable for dial-up retrieval, and thumnail). The Canon SDK forces me to convert all images to TIFF, load the TIFF once and convert it in memory three times. BTW, the EOS10D claims to provide discrete shutter speeds and apertures but the RAW files record intermediate values. Not surprising that the camera has finer TV and AV resolution that the camera UI suggests really but annoying to handle in imaging software.
Note to self: Don't forget to assure the world I'm not in favour of communism the next time I post a "defense" of it
Hear, hear. On paper Star Trek is the story of a communist future. Marx proposed a utopia like that in Star Trek as I understand it. Communism was turned into a dirty word by dictators, not by communists.
...at least from Microsoft perhaps. This sounds like a straight copy of Novell's data protection server feature and, no doubt, a few other technologies from various platforms. I wonder if they'll even copy the Explorer shell integration? I think the Novell one is a bit more than RSync and traditional backup but is limited to Novell's file system (NSS) as I understand it. It's more like CVS meets backup. It is disk to disk but the target disk isn't a duplicate of the source (i.e. not COPY, XCOPY or RSync). The target is a database of versions. I think Microsoft's is similar. In NSS the file system internally tracks changes and can give you them in a list (doesn't need to be trawled for changes). It supports snapshot with copy on write at the file and volume level so you can exclusively lock a file without preventing it from being changed (though you have to have some knowledge of concurrency among your applications to use it safely). Files retain all security attributes in the archive. Users can do their own restores from their workstations (Windows ones get a context menu option) and their access rights are enforced in the retore software so they only see files for restore that they had righs to originally. The common restore case: single file due to user error I believe. If users can restore their own files you don't need to recover a tape, cue it and wait for it to spool to the file in question.
Someone did succeed in defending themselves against criminal charges relating to some use of their PC based on trojan horse software found on their PC. The defence was, IIRC, that the trojan was responsible for the criminal activity. That's what made me think this was an effective get out of jail free card.
My understanding is that civil suits have a lower burden of proof and I believe the RIAA is using the civil process. But my guess is that it would be an argument with a fair chance of success, especially before a jury of average PC users. It would probably come down to whether or not the RIAA could do a hands on analysis of your PCs. If so then they'd find the MP3 files and the playlists that refer to them. You could try to argue away the presence of the files as the hacker's use of your storage but you'd have trouble with the playlists I guess. I have a lawyer friend that I might ask about this.
<excessively pedantic>
Isn't it:
-1 <= sin(x) <= 1
The OP said it wrong I believe by saying "sin(x) can have values between positive 1 and negative 1" and failing to add inclusive. 1 isn't between 1 and -1 and -1 isn't between 1 and -1.
At least since the OP said his/her x was a value between zero and pi and didn't add inclusive then sin(x) is positive, it might have been zero (is that positive or negative) if the OP had said x was between zero and pi inclusive.
</excessively pedantic>
So, just about any law you can break with a computer is now fair game. When you go to court just refer to the three minutes it could have taken some nefarious hacker to use your network without your knowledge. Since the likelihood of such an attack is low then I recommend everyone use a dictionary entry to generate keys. It will keep your neighbours off your network and you'll leave yourself with a perfect reasonable doubt defence when sued or prosecuted.
Is it just me or is the Groom Lake imagery lower resolution than other parts of the country. Also, you can't zoom in as far as you can in other parts of the country. For the spooks watching, I merely tried this in order to be able to report any breach of national security to the appropriate authorities. Now where's that tinfoil hat...
NPR (and NPR "member" stations presumably) and the BBC World service are "members" of PRI as I understand it. Members having the same interpretation as in NPR "member" stations. I frequently hear the phrase "PRI member" referring to individual NPR "member" stations and NPR (particularly the NPR news organisation) used during "The World", a daily PRI show broadcast by NPR "member" stations.
Add to that the BBC (along with NPR, a member of PRI). All of the BBC radio stations have content available for 'Net re-broadcast (I believe you have to record them while playing as well). Radio 4 has excellent speech content with some fine comedy alongside in-depth art, science, current affairs and analysis (e.g. political interviews with members of both sides of an issue in the same studio at the same time).
Some of the BBC music stations are pretty good too. Many of them are segmented by market the way that US radio is but none of them have the sort of motivations that make much of the US radio I have heard just crap (IMO). Long live NPR!
If the other thing from the other company is worth what you pay for it then you pay nothing for the iPod. Arguably buying the other thing from the other company is "authentication" only for the free iPod. You did want that BowFlex, matching luggage and super-sized viagra shaped Rolex watch didn't you?
I'm scared to release the source, though I have thought about it a lot. It's mostly hubris that stops me, plus I'd like to avoid the inevitable abuse about my style and suggestions about how to implement it "properly" ;-) It's hard knowing which kudos to chase: the prospective employer with a million dollar offer for someone that can do this work; or the community. I'm a utopian with a family to feed. As an "anonymous coward" poster you might understand me being a named coward.
The DirectX SDK has some very good Direct Show examples and I used them for the source to real-time display parts of WebCam. There's a little filter builder test tool in the DirectX SDK that lets you chain elements to see what goes together to connect a source to a window.
WMV was an absolute bitch. The examples in the SDK are just rubbish and I ended up doing bizarre things like writing a program to dump the entire state for all of the installed Windows Media profiles in order to find out some of the undocumented properties that are effectively necessary for decent results. The default profiles are fine if you are using "Windows Movie Maker" but not for reasonably configurable applications.
GDI+, OTOH, is just fabulous (to program with). It's pretty well documented and examples aren't really necessary due to its simplicity. Though it's not without its own stupid behaviours, e.g. inability to construct an object to create an EXIF tag (though I haven't looked into whether or not that's just a feature of the .NET framework GDI+ interfaces). One of the other apps on my downloads, Thumber, heavily uses GDI+. That introduced me to another SDK that is just rubbish: the Canon SDK for the RAW image format from its digital SLRs. It's functional but gives very poor access to files.
Anyway, I'd prefer not to publish the source but you can always mail me and I'll discuss the elements of it and address any questions. Send mail to the dmair mailbox in the mair-family dot org domain.
Er, wasn't the point that the useful (intended) doppler shift comes at the cost of another doppler shift in the signal coming from the camera to the viewer. IOW, when there is blue shift that moves the IR into the visible spectrum as the camera is moving towards the subject the camera is flying away from the receiver so fast that there is a red-shift in the signal from it. That was my understanding at least. FWIW, I didn't know if that would affect a cabled webcam or only a wireless one (obviously for a cabled camera you'd run the wire through a collar at the hob of the rotating parts).
...Get one here: http://www.mair-family.org/Downloads/ Geoff says the software he had isn't very good. I couldn't find anything I was prepared to pay for so I wrote my own and publish it at that link. Sorry Windows only and WMV output. You can read the feature list for yourself but, in terms of Geoff's difficulty, it will capture frames at between 4s and one hour per-frame (configurable) and create a 15fps WMV.
Go ahead, /. my web server...Hopefully my ISP is tolerant.
Despite the growth in broadband capacity I am unconvinced that it will have the capacity to support HD video at DVD like quality for multiple homes in a single neighbourhood any time soon. Rotating HD capable media will arrive a long time before that and the players will play your existing DVD collection. Sounds like a hard habit to break. In places like the UK where the wiring infrastructure is ancient it's hard to see how broadband will ever develop sufficiently for complete replacement of physical media (despite the BBC and UK government's plans). Wireless may resolve that but not any time soon and not at all for many homes. This is true for many European, American and African countries and many cities in the USA, all of which can support some form of DVD market. For the studios this means supporting more than one distribution medium versus DVD/future DVD where one disk fits all (region code notwithstanding).
A SMTP proxy may be as good: http://sourceforge.net/projects/assp/ I use this configured to give an error response on delivery of messages it rejects. It receives the whole message first and all spam goes into a dump mailbox. So I get to monitor it's filtering and cause failures on incoming spam. FWIW, I've noticed my daily spam message counts decreasing since I enabled error on spam behaviour.
Segments exist in 32 bit x86 flat memory model environments too. Code is loaded from cs:eip and data is accessed at ds:offset (there are several modes for determining offset). A flat model makes it possible for all references to be near (no explicit segment) but that only means cs is implied (code) or ds is implied (data) es may be implied too for string instructions. The OS loads cs once and ds(,es,fs,gs) once for any user mode process but segments are still loaded and used. NB: I'm talking about what you can do (in response to you your "could not" point), not what you might prefer to do when implementing an OS (I don't like segmentation either). I don't believe it is true that you can't make memory readable and not executable but AFAIK these exploits depend on writing and executing anyway. I believe you can make addresses executable but not writable on an x86 without NX. Segmentation allows you specify a base address and limit for a descriptor that can be selected by cs (what eip must be an offset in). If you have no writable data descriptors that overlap any code descriptors then a user mode application can't execute arbitrary code by poking instructions into memory. IIRC, you'd get a GPF trying to write opcodes where you could execute them or a GPF attempting to jump to an address where you could write opcodes. You can also make code readable by putting it in read only pages but, if cs has a 4GB limit, you can't prevent a program from running arbitrary code it pokes into memory it can write to. I think that is the real cause of the problem and the OS chooses the values of the base and limit for cs. NX provides executable control in the paging system but it has been available in the segmentation system since at least the 386 (can't remember 286 descriptors). You have to use segmentation even in a flat model (implicitly) but you don't have to use paging. That (and MS bugs) may be a factor in the time it has taken to develop NX. FWIW, I don't advocate segmentation and an executable page attribute has been missing from the x86 MMU since the 386. IMO it is an omission in the x86 design that you can have supervisor and user segments in read, read/write and executable types, but supervisor and user pages in only read and read/write types.
I don't actually know how they measured it (and I'm not an astrophysicist and don't offer this as opinion, let alone fact) but 1.5 million miles per hour probably creates a significant doppler shift. The spectra of stars shows distinct bands that correspond with the absorption of photons at specific wavelengths. The reason the absorption is at specific, discrete wavelengths is that electrons in specific elements (e.g. Hydrogen, Helium, etc) absorb at specific wavelengths. The distance between absorption lines for different elements is fixed. For example, say Helium absorbs at some wavelength x and Hydrogen at some wavelength y. The difference between x and y is a constant (even under a doppler shift). So, if you see two absorption lines and they are x - y apart then one is the Hydrogen and one the Helium. This is overly simplified but the principle is the same (it just involves more than two lines). You can then tell how much doppler shift because you know what wavelength they should appear at and see what wavelength they actually appear at. Doppler shift is directly proportional to speed (I hope that's accurate). So, knowing the doppler shift you know the speed relative to the point of observation. Presumably you can work out the speed of the earth relative to some object by measuring the relationship between doppler shift from that body and the doppler shift from various other celestial bodies. Or, maybe they had a very long tape measure, a very powerful radar, or a very powerful laser or just guessed ;-)
The Boeing computers auto-tune multiple VOR receivers (actually uses DME as I understand it) to correct INS drift. If you are receiving from two VORs then you can plot an arc of possible positions, three and you can plot a point in space. Then it's only a matter of accuracy in the system. If you look at the Boeing ND with stations displayed then the auto-tuned VORs are highlighted in blue.
Spoke to my CFII/ATP friend. WAAS (augmented GPS) approaches are appearing regularly. While they are precision approaches none of them are zero visibility qualified. They are about 10m vertical less accurate than equivalent ILS approaches (that's a 20 foot minimum threshold crossing versus 50 for an ILS) and there is a consequential limitation on glideslope angle that makes zero visibility landings impractical (steeper approach, flare up to 30' too high or too low). The ones we looked at were all slightly worse minimums than an equivalent ILS approach (not Cat III), for the reason above presumably. I can only remember the RVR was 1.5 miles but I can't remember the DH, I think it was 500'. I guess this doesn't prevent them from being used for auto-land however (you can always hand fly once you can see the runway). My friend tells me that the FAA is actively giving vibes that the status of WAAS GPS approaches are not going to be improved upon (period). I get the impression that the FAA doesn't trust GPS unless you can see the things you might hit.
Radio altimeters, yes. The 747-400 will talk to you. If you want it to it will call out the reference speeds on take-off and useful altitudes on approach. It also tells you when it isn't happy with your flying! Listening to a 747 barking "GLIDE-SLOPE, GLIDE-SLOPE" then a minute later "TERRAIN, TERRAIN, PULL-UP" because it is unhappy with the radio altimeter reading makes for really sweaty palms.
Aircraft with engine out (including gliders) always have right of way (I think that non-steerable aircraft, e.g. balloons, maybe have a higher right of way). I once had to call another aircraft on approach to get them out of my way when I was engine out and my instructor had to go on the radio to insist that they comply with the regulation they obviously didn't know. Helicopters look like fun but are really expensive.
Just out of curiosity, is the Tacoma approach in question WAAS-only, or just IFR-GPS?
It's an old chart so it is VOR-GPS only.
I also don't follow the logic about the reference point being between the two runways. If the computer knows where it is, and where it's going, any 'reference point' (which is actually provided by at least 4 sats possibly plus a ground station) is completely arbitrary. And if the autopilot can't (physically) fly it GPS, it can't fly it ILS, either, right?
Yes and no I guess. The published procedure is what the crew/aircraft is supposed to follow. For the VOR-GPS plate I looked at the GPS approach is a GPS overlay of the VOR approach. The approach is a runway heading to the VOR but the VOR is not on either runway, it's between the two (nearer the right from what I can tell). So, the published procedure is to fly down to that VOR on the runway heading but to the right of the left runway or to the left of the right runway. Then, when the field is in sight, side step to the correct centreline and perform a visual landing. That appears to make it largely a locating procedure because you may as well line up on the correct runway as soon as you are able (gets you stabilised as early as possible). There's no guided profile (glideslope) so the approach is officially non-precision.
I once had the good fortune to get a few hours on a 747-400 Level D simulator (a BA 747 training captain needed help with his laptop). I got to watch two first officers doing a dry run of the check-ride they were doing the next day. It was a wild set of coincidences they were given but I got to see them both do a non-precision NDB approach into Gatwick. The ceiling was 600 feet AGL. The pilot flying was calling the distance to run every mile using the navigation display and the pilot not-flying would call the altitude for an on profile approach. It was obvious to me from the ND that the aircraft could have handled the approach better using the INS and VOR augmentation than either pilot (particularly one of them) was able to by hand flying using the ND but regulation and policy prohibited it. The aircraft is only pre-programmed with the published procedures (plus it can probably handle straight in using the ILS without needing a procedure - select the ILS frequency, select the auto-pilot in approach mode and sit back). So, if you want to use the auto-pilot for an approach and landing your only options (for the types I've had any kind of access to) are to hand program the approach waypoints (too slow and prone to error plus, at least the 747-400, won't auto-flare) or to select a published procedure from the flight computer. If the only published GPS procedure is a non-precision approach (no glideslope) or has no guidance onto the runway (the VOR-GPS at Tacoma for example) then regulations (and probably company policy and auto-pilot implementation) will prevent you from using it as a zero visibility approach. FWIW, company policies often enforce stricter limits than those enforced by the FAA. Incidentally, cool 747-400 software bug... The BA pilot that let me fly the sim told me he once did a re-positioning flight to move a 744 from Gatwick to Heathrow (less than 10 minutes flying time). He tried to tune the Heathrow ILS frequency on the radio only to discover you couldn't select an ILS within 10 minutes of a take-off. He assured me it's been fixed.
I know that typical GPS approaches have worse minimums than comparable ILSs, as the (unaugmented) accuracy is not as good, particularly when it comes to altitude. I'd have to check, but IIRC, WAAS is supposed to be accurate to within 1.5 meters. And believe me, I (and many other students) have dropped a plane out of the sky from higher than that. :)
I did glider training once. Flying a three degree glideslope is too risky so you do ten degrees or something equally wild. It's about the best roller coaster I've ever ridden. My instructor (the friend I've been paraphrasing) used to shout at me f
I'm currently looking at the Cat III ILS RWY 16R approach chart for Seattle Tacoma. The Straight in ILS 16R (S-ILS 16R) minimums for CAT IIIc (any aircraft category) is N/A (i.e. no runway visual range limit - i.e. zero visibility). The Cat IIIb has a RVR limit of 300 feet. None of the listed Cat III minimums includes a decision height. The following is written on the chart: "CATEGORY III SPECIAL AUTOLAND EVALUATION REQUIRED: "CATEGORY III ILS-SPECIAL AIRCREW & AIRCRAFT CERTIFICATION REQUIRED" My understanding is that a Cat IIIc approach and landing is a coupled approach only and approach and landing are permitted in zero visibility. I also checked some Cat III charts for Denver and Salt Lake City. All have no RVR limit or decision height for a Cat IIIc approach. All have the same certification notice. Most of the runways in question have multiple ILS approach charts, several non-cat III plates and one Cat III plate. The GPS approach I looked at, Tacoma GPS/VOR 34L/R for example, has a lowest decision height of 840 feet and RVR of 240 feet. The Tacoma GPS approach I looked is for both 34L and 34R so it has to require a visual final and landing since the reference point for the final approach course is between the two runways. Even with better accuracy than is currently available from GPS an autopilot couldn't fly that to landing.