Wikipedia:Reference desk/Archives/Computing/2015 July 12

= July 12 =

Only 3 operating systems
After some observations, i have noticed that there are only 3 types of operating system: Mac OS, Linux and Windows. My question is: Are there more types not based on these 3? I know iOS is Mac OS. Possibly for other computers? RocketMaster (talk) 03:35, 12 July 2015 (UTC)


 * See Operating system and List of operating systems. It's true that most modern operating systems on microcomputers are descendants of or heavily influenced by Unix (this includes Android, iOS, OS X and Linux), with the major exception to this being Windows (and ideas from Unix have influenced that too). On mainframes, Unix derivatives are also common, but there are still proprietary operating systems in common use on some manufacturers' computers.- gadfium 03:54, 12 July 2015 (UTC)


 * There was OS/2. StuRat (talk) 04:22, 12 July 2015 (UTC)


 * And many others such as DOS, CP/M, the Apple II system, others in microcomputers and many for minicomputers and mainframes. Bubba73 You talkin' to me? 04:27, 12 July 2015 (UTC)


 * You're fixating on desktop computer operating systems - and only on the current popular ones! There are many more operating systems suitable for personal computers, and many more have existed historically (even if they are uncommon today).  If we consider other types of computer systems - servers, microcontrollers, distributed systems, video-game consoles, application-specific software - there are many more operating systems.
 * Start by reading list of operating systems. There are hundreds of them in that list!
 * You might not be familiar with these softwares as operating systems, at least not by brand name; but surely you've seen a video game console, or an airplane, an ATM, a digital motion picture projector at a movie theater, or a microwave oven. All of these devices commonly have computers inside of them.  Many of these computers run an operating system, and that does not necessarily need to be Windows, not Linux, and not OS X.
 * Nimur (talk) 08:25, 12 July 2015 (UTC)
 * Just as an interesting FIY, almost all ATMs do actually just run on windows. Vespine (talk) 23:45, 12 July 2015 (UTC)
 * You are absolutely correct: it is true today that most ATMs you will encounter out in the wild are probably running Windows XP or one of the newer Embedded Windows softwares. There was a time when almost all ATMs ran OS/2, but this has probably not been true for the last decade.  Modern ATMs were probably a poor example to demonstrate my point!  Nimur (talk) 05:37, 13 July 2015 (UTC)
 * Specifically, the CEN/XFS client-server architecture for financial applications on the Microsoft Windows platform dominates, although Linux is making inroads.


 * http://www.telegraph.co.uk/technology/internet-security/10543850/Hackers-target-cash-machines-with-USB-sticks.html may be of interest. --Guy Macon (talk) 07:24, 13 July 2015 (UTC)

Deleting photos from Gmail's people widget
I have long been aware that if I upload an attachment to gmail, I am uploading an attachment to google. That has never been a problem.

But I have recently received private images from third parties not aware of google policies, and the private images they have sent me started appearing under their name on the right side of my browser when I read their emails. Some are images I don't like or, much more seriously, which the sender would not want shared or displayed. I understand one can turn off the display of the people widget. From googling I see no way to tell google to erase those images that belong to parties who did not give their consent for the image's use. I understand that if you even delete the email entirely, the image remains.

Is there any simple way to get these images removed from google's servers? Is there any regulatory agency or published legal action of relevance in the United States? Thanks. μηδείς (talk) 03:50, 12 July 2015 (UTC)
 * Removing Content from Google: Legal Help. You should consult an attorney, because this process is neither easy nor risk-free.
 * Nimur (talk) 15:37, 12 July 2015 (UTC)


 * Yes, thanks, I was afraid of that. It looks like I'll have to await regulatory action or a class-action lawsuit. μηδείς (talk) 17:25, 12 July 2015 (UTC)


 * The emails are appearing in your browser when you read your own emails, is that correct? It seems to me very likely that Gmail is displaying those pictures to you because they exist in your own archived messages. That doesn't sound like a privacy breach to me. You were the person the images were sent to and you are the one seeing them.
 * When you say "I understand that if you even delete the email entirely, the image remains", is this something you have tried? I can't see where the picture could linger after the message containing it was deleted. I suppose it's possible that a thumbnail version, created automatically for easy previewing, could hang around in a cache for a short time. —Noiratsi (talk) 18:12, 12 July 2015 (UTC)
 * As with many companies, I'm fairly sure simply deleting the emails will not remove them from the servers immediately, perhaps that's what Medeis is referring to. Also it's possible the emails will end up somewhere else, e.g. it may be relatively easy to add them to Google Drive from Gmail. And it's possible that if they end up as thumbnails for people, the thumbnails will get detached from the emails and will need to be removed seperately. Generally though, I agree that deleting the emails will remove them from Google's servers eventually. I don't see that following the DMCA process suggested by Nimur is helpful here when the only reason Google is keeping copies is because Medeis is keeping them and Medeis is the one who wants them deleted. Nil Einne (talk) 21:57, 15 July 2015 (UTC)

Changng PDFs into black and white
How can I change my color PDFs into b/w to cut the file size down?--86.176.9.176 (talk) 14:40, 12 July 2015 (UTC)


 * One obvious way is to print it to a PDF writer and select black and white in the output options. I don't know how much in size the file would be reduced. --TrogWoolley (talk) 16:02, 12 July 2015 (UTC)


 * If you can run Ghostscript ("gs") from the command line, you can do:

gs -q -sDEVICE=pdfwrite -sColorConversionStrategy=Gray -o OUTNAME FILENAME
 * You may also need to specify a page size (e.g. ) for the output file.
 * See here, and in particular here, here, here, and here. --70.49.171.162 (talk) 18:54, 12 July 2015 (UTC)


 * It's unlikely to save much space. Removing the color from text and vector graphics will have almost no effect on the size. Bitmapped images are likely to be encoded as JPEG or JPEG 2000, which encode color pretty efficiently (usually it's less than 1/3 of the total image size). -- BenRG (talk) 19:57, 13 July 2015 (UTC)


 * One issue that comes up when changing color files to black and white is what precisely you mean. For example, if you have yellow text on a white background, do you want that text to become a very light grey (having the same darkness as yellow) on a white background, or do you want it black on white, to be more visible ? StuRat (talk) 22:00, 12 July 2015 (UTC)


 * Even worse, what happens when the top half is orange text on a white background and the bottom half is black text on an orange background? --Guy Macon (talk) 07:28, 13 July 2015 (UTC)


 * I assume that the question wants to make the PDF grayscale, not black and white. Black and white is technically just two colors: completely black or completely white. However, most people say "black and white" when they mean "grayscale." If the question is really about grayscale, it appears that it is incorrectly assuming that changing color to grayscale will reduce file size. Grayscale is not smaller than color. If I have 8 bits for color, I use all 8 bits for color. If I have 8 bits for grayscale, I use 8 bits for grayscale. You can convert, say, 32-bit color to 8-bit grayscale. You could just as well convert 32-bit color to 8-bit color. Changing bit depth is a separate process from changing color type. You could also change 32-bit grayscale to 8-bit color to reduce file size. 199.15.144.250 (talk) 12:19, 13 July 2015 (UTC)


 * There's also black and white dithering (using just black dots, but spacing the black dots out to achieve the appearance of grays). That looks worst of all.


 * 32-bit greyscale seems like excess bit depth, compared to 8-bit, for most applications, like a photo. 32-bit color, on the other hand, is a rather noticeable improvement over 8-bit color.  Also note that under the HSL and HSV color spaces, only the L (lightness) or V (value) need be retained when converting to greyscale.   StuRat (talk) 13:13, 13 July 2015 (UTC)


 * Our MRI produces 32-bit grayscale and there are still times that they have to go back and repeat the process on a specific area because the image becomes blurred when zooming in. The bit depth is all based on need. I'm sure newer MRI machines use 64-bit grayscale to avoid repeat scans. 199.15.144.250 (talk) 14:18, 13 July 2015 (UTC)


 * At 32 bits, there should be 232 shades of grey, which is over 4 billion. I can't imagine that the MRI looks blurry due to 4 billion shades of grey not being enough (and "blurry" isn't how one would describe insufficient color depth, in any case).  It must be blurry for some other reason, like the patient moving during the scan. In fact, if the 32 bits was the problem, then rescanning again at 32 bits wouldn't make it any better. StuRat (talk) 15:09, 13 July 2015 (UTC)


 * In tractography (is there an article... yes!) mapping 2D images to 3D requires immense bit depth. I used "blur" to avoid getting technical about artifacts such as barring, striping, or starring. Similar effects appear in dithering. Scans are not always done on the entire body. You can mechanically zoom in on a specific area as opposed to doing a software zoom, which is where the interpolation artifacts show up. I don't work over in the CT lab, but I assume they have the same issues that MRI has. 199.15.144.250 (talk) 15:28, 13 July 2015 (UTC)


 * Yea, that's the same issue with digital zoom versus optical zoom. Digital zoom is just about useless, since the resolution (pixels per inch or cm) just isn't there, and you can't create missing information with software.


 * Re: "mapping 2D images to 3D requires immense bit depth", I don't think that's quite the right name for it. It may require lots of working memory (RAM), but that isn't really bit depth.  For example, if you wanted to represent 8 layers as a single image, you could have eight 8-bit pics all combined into one by making each pixel 64-bits, but that's not the same as a single 64-bit pixel.  StuRat (talk) 15:36, 13 July 2015 (UTC)


 * This is quickly getting very off topic... The shoulddistance (void) between layers is where interpolation takes place. Please note that I do not work on the MRI software, but I do listen to the engineers. It is not as simple as interpolating between point X/Y on one image and point X/Y on another. You have to interpolate between the vectors of interpolation as well. This is where artifacts appear. One way to fix the problem is to take more images with less void - which is more expensive and makes the entire scan take a lot longer. Another way to fix the problem is to scan at a higher bit depth. I never really cared for the mathematical reason it works. My focus is on trying to find a way to explain to people that when you a logarithmic increase in storage needs because your MRI machine is producing much larger images, linear growth in data storage machines isn't going to work. Management literally (yes, literally) gave me the "it took 15 years to fill 50TB of space, so we should be good with 100TB for another 15 years." 199.15.144.250 (talk) 15:58, 13 July 2015 (UTC)


 * Interpolation seems like a bad thing in the context of medical scans, since it makes it look like you have better resolution than you really do. So, if there's a medical problem in the void area, which doesn't extend onto either adjacent layer, the interpolation will show a layer in the void with no problem, which is a dangerous lie.  Better to show nothing there than to show that. StuRat (talk) 20:09, 13 July 2015 (UTC)

Returning to 199.15.144.250's point, it's correct that converting color to grayscale will not necessarily reduce the file size. However, if the PDF contains, for example, JPG images with 24-bit color, then reducing these to 8-bit grayscale may well indeed reduce the file. Before posting the gs command line above, I tested it on a copy of the 191-page rail accident report (the official report on the Lac-Mégantic disaster) that I happened to have on my computer, which contains a number of color photos. The file size shrank from 2,987,415 bytes to 2,486,924. --174.88.133.35 (talk) 04:35, 14 July 2015 (UTC)


 * OP here. Original documents were black text on white b/g. They have been scanned in color giving a very large file size. I want to change them back to black and white (not grayscale). How do I alter the PDF to do that?--86.176.9.176 (talk) 17:04, 14 July 2015 (UTC)


 * I found a way to do it, and the resulting file was 43 times smaller, but it uses a screen grab, so only works for one screen's worth at a time. I did a screen grab with the Print Screen button, pasted into Microsoft Paint, did a Save As, selected monochrome bitmap as the Save As Type, saved it, opened it using GIMP 2, then exported it as a PDF file.  There has to be a better way to do this, but this is a "proof of concept", that a true black and white (monochrome) PDF file can be much smaller than the color file it was created from.  (I also tried using Adobe Reader, and, while it doesn't appear to have an option to convert to monochrome, perhaps the more complete Adobe Acrobat does.) StuRat (talk) 21:32, 14 July 2015 (UTC)


 * A screen grab will give you an upsampled or downsampled version of the original image, which will worsen the quality (and the file size, if it's upsampled).
 * You can use pdfimages to extract the images in their original resolution (there are Windows and Mac OS binaries here), then fix them in any image editing program and re-export to PDF. ImageMagick will likely also work, but I've never used it. -- BenRG (talk) 03:19, 15 July 2015 (UTC)