Wikipedia:Reference desk/Archives/Computing/2015 August 13

Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

August 13

Which Social Media Channels Should I Use For My Business?

Here is a list of the most popular and effective social media channels: ... biggest social media channel, with an estimate of 1.11 billion people using <spam link redacted> the site .... You can share content, find answers, post and view jobs, make business contacts — Preceding unsigned comment added by AlfredNobell (talk • contribs) 09:18, 13 August 2015 (UTC)

I've removed your link about buying instagram followers. The Wikipedia reference desk is for asking factual questions, not advertisting your own business. -- Finlay McWalterᚠTalk 09:28, 13 August 2015 (UTC)

Android app to count cells in micrograph images?

Hi. I need to count cells in images and there are a lot of cells in a lot of images to count. The images come in pairs and I have to count an area (usually the upper left hand corner of the image and then count the number of bright yellow ones in the corresponding image. I usually do this by putting block dots on the images in the GIMP (GNU image manipulation program - like Photoshop but free and not quite as powerful) and using an app on my phone to count. The black dots stop me from counting the same cells twice. I'd like a way to do this on my touch screen tablet because it would be a lot faster than moving the mouse around. Is there a way I can put black dots on these images (preferably on an invisible layer that I can switch on and off) that will also count the number of times I press/place a dot?

Bright field micrograph

Fluorescent micrograph — Preceding unsigned comment added by 129.215.47.59 (talk) 10:32, 13 August 2015 (UTC)

So, you're counting manually, but you use software to help you keep track? I recommends using software to actually automagically count the cells. This is a common technique in many labs, but I have not set it up myself. ImageJ and the Fiji_(software) based on it are both free and up to the task. Fiji doesn't currently work on Android, but it might some day [1]. Here is a plugin specifically for counting cells [2], and here is another description of cell counting methods [3], one of which is also a manual process with computer assistance, but seems more streamlined and faster than what you are doing. I guess my point is that I would consider doing this by algorithm on a desktop or laptop rather than by hand on a tablet. It will be faster and more reliable. SemanticMantis (talk) 14:10, 13 August 2015 (UTC)

One other idea more along the lines of your specific question: consider Inkscape, available on Android (along with the GIMP, e.g. here [4]). One way you could do it - import the raster image to Inkscape, and set it as a background. Then place small circles over the cells (they will be vector graphics), using a different layer (this will be toggle-able as you desire). When you are done and have a final .svg, you can programmatically count the circles using any number of XML tools (conceptually, use grep to count the number of "circle" elements, but you'll have to figure out some of the details). This actually seems harder to me than getting Fiji working on a desktop, but it is more in the spirit of your question. Let us know any of these options works for you, or if you have further questions. (p.s. does User:Wnt) have any comment on this?) SemanticMantis (talk) 00:40, 14 August 2015 (UTC)

Sorry, you already said what I was going to say and a lot more. I'm afraid I don't even get why the OP is trying to use a phone to do this - maybe I'm behind the times, but I have this archaic notion you want to do image analysis on something with a really big screen. Searching cell counter app does deliver a lot of hits though, many for Android. It occurs to me that it would be cool to set up a web site that you can upload an image and get the counting done server side "for your convenience" (with insertion of some highly targeted ads...) Wnt (talk) 00:49, 14 August 2015 (UTC)

I think OP was using a cell phone much like a Tally_counter, or I guess a type of Counter_(digital), whilst doing manual GIMP stuff on a bigger computer, but it's a little unclear As for your website notion- that is probably a decent idea to provide a useful service and maybe make a little money. Unfortunately I don't have the stomach for the kind of QA/QC that would require for scientific usefulness (or maybe not -"no warranty expressed or implied" and all that). I suspect there's either some grant money or business profit in such a venture, so maybe we should hatch a plan ;) SemanticMantis (talk) 01:25, 14 August 2015 (UTC)

How to print from Blackboard

I am a correspondence student at UND and UND is having a blackboard shutdown this weekend. I really want to get ahead, so I am trying to print from blackboard. When I try to print on my Mac, page preview only shows one page cut off, it does not include all the information from the blackboard web page. I can copy it in to Word and print there but I lose much of the formatting, which includes things like charts and tables.

Is there anyway I can get a whole blackboard page to print without it being cut off? Is there a print button within blackboard? --2602:30A:C019:2920:FDE7:B0BE:195C:CB20 (talk) 14:03, 13 August 2015 (UTC)

It sounds like the first type of print just does a screen grab, in which case you can scroll down and do a screen grab for each page. Not an ideal solution, but you do at least preserve the formatting. StuRat (talk) 14:50, 13 August 2015 (UTC)

I use three versions of Blackboard at different universities. None have a print option. Firefox tends to do a print page a little better than IE. You can also right-click in Firefox and print a frame instead of the whole page. I don't know if IE can do that yet. 209.149.114.32 (talk) 16:43, 13 August 2015 (UTC)

Hi. I am a Blackboard user to. I use CutePDF. CutePDF will print the entire page. Although it will break it up into multiple pages. What it is is its a virtual printer driver that takes the print its given and converts it to a PDF file and allows you to save it to your desktop(or where every you want). You use it by simply pressing CTRL+P to tell firefox or IE to print the page and in the printers selection box select the one that says CutePDF in its name. Next click print and wait. You should see a save file as box pop up that allows you to name and save your new PDF file. Note:if you end up getting it, please be sure to carefully read the installer pages. As for if you breeze through, it will install adware on your PC. So just read the pages in the installer carefully and dis-select any toolbars or extra apps that it wants to install. Hope this helps, —SGA314 I am not available on weekends (talk) 18:31, 13 August 2015 (UTC)

Since OP is on a Mac, CutePDF might not be necessary, as Macs have built-in print-to-pdf functionality. Firefox>File>Print>PDF works for many situations like this. You can even open the pdf in Preview to check it out before you save or print, and note this is (annoyingly) different from the standard "print preview" options. SemanticMantis (talk) 20:09, 13 August 2015 (UTC)

Yep, that functionality wasn't working. Anyways, after being stuck quite awhile, I decided the best option was to Cmd+S the page and print the page source, skipping the first page which is just Blackboard buttons and the like. Not ideal, but I was able to make it work. Thanks for the help.2602:30A:C019:2920:FDE7:B0BE:195C:CB20 (talk) 22:47, 13 August 2015 (UTC)

published operating system source code

(A) Which operating systems have their source code published in book form? (Uncle G casually mentioned three of them at Wikipedia:Articles for deletion/MicroC/OS-II).

(B) Which Wikipedia category or article would be most appropriate for listing all such operating systems that we know about?

--DavidCary (talk) 17:43, 13 August 2015 (UTC)

Well Windows certainly wouldn't be one of those. —SGA314 I am not available on weekends (talk) 18:22, 13 August 2015 (UTC)

First, the ISBN he listed is invalid. I assume he is referring to "Operating Systems Design and Implementation." It does contain some source code for Minux and XINU. It does not contain the entire operating system. Second, you must define what an operating system is (there is no widely accepted definition). If you allow a minimum operating system, you can write one in a few lines of code that just says "Hello" and does nothing else. That can be in a book. Finally, what is the notability of having source code in book form? Is Windows non-notable because the source code is not in book form? 209.149.114.32 (talk) 18:23, 13 August 2015 (UTC)

Primarily as a point of interest, I'll note that there was a published source code listing for the Atari 800 (and 400); see this PDF. That "tech ref" was of great interest to a younger LarryMac, and provided a great glimpse into the inner workings of a "modern" computer of the time. --LarryMac | Talk 19:14, 13 August 2015 (UTC)

For question (a): If we concern ourselves *only* with the kernel, back in 2004, the Linux OS was already 4.3 million lines of code. The number of lines per page in a typical paperback book varies between 25 and 50 - with a typical value of 35 lines. So the 2004 Linux kernel would require around 123,000 pages - which would be around 1,000 volumes...it's probably a lot larger now.

Windows 7 was estimated to be around 40 million lines - but it's harder to say what's "Kernel" and what's the rest of the stuff...but it's going to be at least as much as Linux.

No 'modern' desktop operating system is going to fit into a single book...so what about cellphone operating systems?

Android is around 3 million lines.

The Blackberry phone uses a "microkernel" approach - and it's kernel is a mere 10,000 lines - which would fit handily into a paperback book...but as soon as you want to do anything with it, you'll need a bunch more components - and then you're back looking at code sizes similar to Linux/Android.

SteveBaker (talk) 23:33, 13 August 2015 (UTC)

The entire source code of the Version 6 Unix kernel was famously published (originally illicitly) in the Lions Book, along with commentary, as a teaching tool. Of course this is 1970s code, so it's not exactly a modern operating system (the given code only runs on the PDP-11), but it's definitely of historical and pedagogical interest. The original IBM PC included the complete source code for its BIOS in its manual. Outside of operating systems, the entire source code for PGP was published in 1995 in book form, to evade U.S. laws restricting the export of cryptographic software. In a similar vein, the Electronic Frontier Foundation published the source code to its EFF DES cracker in a book. Now, these examples illustrate why source code is pretty never published in book form anymore: there's just no real demand for it. As the older examples show, some source code was published in ye olde days, when there was no Internet to distribute software easily, and when software tended to be less complex. The only more recent examples were hacks of the legal system. --108.38.204.15 (talk) 04:21, 14 August 2015 (UTC)

Yeah the v6 UNIX kernel may well be the only true example...and it's not exactly "modern". It was famously possible to print it out on fanfold paper and store it in a briefcase...along with the shell and all of the command-line tools that came with it. Small 'example' operating systems might fit into a book - but I doubt that anything that people commonly use on desktop computers in the last 10 years would qualify.

The IBM PC BIOS isn't an operating system - it was (essentially) just the keyboard, screen and disk drivers - plus the boot-loader for the PC-DOS operating system. The BIOS sources for CP/M were also published (and is almost identical to the IBM PC BIOS of the time)...that was the way you got CP/M running on a new computer...you read the 'standard' BIOS code, and wrote something for your computer that had the same interface.

SteveBaker (talk) 17:06, 14 August 2015 (UTC)

What is note 0 in MIDI?

Resolved

Hello everyone. What is note 0 in MIDI? For example, MIDI note 62 is D4. What would MIDI note 0 be? I would think it would be C0(I think this would be the most likely answer) or A0. Thanks for your help in advance. —SGA314 I am not available on weekends (talk) 18:21, 13 August 2015 (UTC)

It is a very low C - one octave down from C0, often written as C-1. Just keep going down from a note that you know. It shouldn't be used in anything though. 21 (A0) should be the lower limit of any real music. 209.149.114.32 (talk) 18:29, 13 August 2015 (UTC)

Cool thanks. I am only needing this for programming purposes. —SGA314 I am not available on weekends (talk) 18:34, 13 August 2015 (UTC)

Google Chrome Issue

Guys, my 'google chrome' 'window' doesn't appear no matter how many times I click on it... 'Opera' works however; the 'window' appears. Any idea how I could get it back to working... -- Space Ghost (talk) 18:50, 13 August 2015 (UTC)

Restart the computer. If you already did this sorry, but you didn't saw you did and history has shown it's a mistake to assume stuff even if it seems obvious. Nil Einne (talk) 19:25, 13 August 2015 (UTC)

I have, many times, it just doesn't wish to open. I'm logged in it with my 'Google ID', I have all my 'boobkmarks' in it... I don't know what to do now. I'm kind of stuck... -- Space Ghost (talk) 20:48, 13 August 2015 (UTC)

Perhaps you should uninstall and re-install Chrome?188.247.76.211 (talk) 05:35, 14 August 2015 (UTC)

Clear caches by pressing CTRL+SHIFT+DEL in browser window. Note: When clearing cookies, You get logged of from Wikipedia an other sessions. --Hans Haase (有问题吗) 18:16, 14 August 2015 (UTC)

The browser window did not appear regardless the amount of times I tried to open it. Anyway, I done what I was afraid of, i.e., reinstalled it... Thanks friends. -- Space Ghost (talk) 18:21, 14 August 2015 (UTC)

Maybe a fault user profile makes the browser fail to start. Kill the user profile after backup. --Hans Haase (有问题吗) 20:29, 17 August 2015 (UTC)

new computer and dual-channel memory

My last few computers have had dual-channel memory where you should add memory sticks in matching pairs for best performance. I just ordered an HP Envy 750 with 12GB of RAM - one 8GM module and one 4GB module. Is this going to hurt the dual-channel memory performance, or does this motherboard even have dual-channel memory? Bubba73 ^{You talkin' to me?} 19:20, 13 August 2015 (UTC)

Well most old dual channel memory controllers didn't support such configs at all, so you'd have no dual channel with such a config. But Intel has supported flex mode for a very long time now, I believe with the Intel P35).

It would have an effect in a memory benchmark, but very few real world applications actually function like a memory benchmarks. Many tests show the effect of dual channel still isn't that great in most more real world scenarios. See for example [5].

If you're using the iGPU and demanding a lot from it (i.e. gaming or certain GPGPU work), it does tend to have a bigger effect, see for example [6] [7]. Since I presume you're using flex memory rather than single channel, the effect won't be so great, at a random estimation perhaps 10-20% performance difference for the GPU.

P.S. If this is a new computer, the motherboard is mostly irrelevant since all Intel and AMD computers haved used IMCs for a long time now.

Nil Einne (talk) 19:55, 13 August 2015 (UTC)

This is a new computer (I ordered it yesterday). Some of the things I do are very memory intensive. It will take 32GB, so I'm going to add two 8GB sticks for sure, and I'm wondering if I should replace the 4GB stick with an 8GB stick while I'm at it. The 3x8+4 = 28GB will be enough, but I'm more concerned about memory performance. Bubba73 ^{You talkin' to me?} 20:11, 13 August 2015 (UTC)

You could benchmark it if you want, but as I've said, very few real world applications on the CPU get much benefit from even dual channel over single channel (which is likely to be a fair amount slower than flexi, particularly flexi with 8*3+1*4). Note that memory intensive is fairly unclear. It doesn't matter so much how much memory your programs use, but whether they are write, reading or copying from memory a lot. You can see in the earlier link, the only program with a significant difference that wasn't just a memory benchmark was Euler3D performing CFD. So if you're doing scientific simulation, perhaps it's worth considering, although even then the results on dual vs single were only 17.9 so on your flexi will probably only be about half that or something. (While the Adobe Premier difference may be noticable, a 6% difference is really still a very small difference, and a 3% or whatever really starts to become insignificant. So for something like video or photo editing, it's IMO not really worth worrying about.) In fact, even the completely synthetic memory benchmark is less than I expected. Nil Einne (talk) 23:11, 13 August 2015 (UTC)

I write some math-intensive programs, some of which do a lot of reading and writing to memory, so they aren't typical programs. For instance, I can often make the program run faster by having it execute MORE operations if I can get the main part of data in the L3 cache. (A 3% overall difference in execution speed can cut a few days off the run time.)

Anyhow, I know I'm going to add two 8GB sticks. I was considering replacing the 4GB stick with an 8GB one at the same time. Since I asked the question, I found out that I can pull out the 12GB of memory with CAS latency of 11 and replace it with 32GB of CAS latency of 9, for only a little more than three sticks of 8GB, but I don't know if that will matter. Does the CL depend on the motherboard? If I can get things out of memory 11/9 faster, it will make a measurable difference on some things. Bubba73 ^{You talkin' to me?} 23:39, 13 August 2015 (UTC)

(Ok, I'll bite: what kinds of stuff are you doing that takes >100 days to run on a nice new fast computer like this? I mean I know some things can take that long (e.g. high resolution global circulation models) but usually researchers would run that stuff on a nice big cluster, ideally one managed and maintained by someone else :) SemanticMantis (talk)

CAS latency, especially in new RAM is not at all a linear relationship. It is definitely NOT the case that some thousand operations on cas 11 RAM that took 11 seconds, will take only 9 seconds on CAS 9 RAM. In fact "real world" difference is mostly non existent since modern RAM architecture uses multiple pipelines, so instructions can be run "during" the cas latency of previous instructions, making CAS latency close to irrelevant in most applications. CAS latency has more details, especially the section on effect on memory access speed. Vespine (talk) 00:56, 14 August 2015 (UTC)

Thank you I'd read that article but not understood it well enough. Bubba73 ^{You talkin' to me?} 02:10, 14 August 2015 (UTC)

The CAS latency article does say "Generally the lower the CL, the better".

I wasn't disputing the general claim that lower CL is "better". I was disputing the specific claim that you will notice any real world difference between otherwise identical RAM one being CL11 and the other being CL9, even with long running and complex operations. Vespine (talk) 05:56, 14 August 2015 (UTC)

Well, I have ordered three 8GB sticks of CL=11 that Crucial guarantees will work. But what does "better" mean in this context, if it doesn't mean faster? Some of these things have several gigabytes of data in RAM and there is only one line of code in the inner loop, so it is doing a huge number of memory accesses - maybe on the order of 10^15 over the entire run (top-of-my-head estimate). Bubba73 ^{You talkin' to me?} 06:43, 14 August 2015 (UTC)

To reply to SemanticMantis - I'm just a person doing the best I can with what I have. There are plenty of open-ended problems in number theory and combinatorics that can run until progress gets too slow, or you get tired of waiting and want to do something else. Bubba73 ^{You talkin' to me?} 01:02, 14 August 2015 (UTC)

Agree with Vespine that stuff that takes over 100 days to run is very rare. But even when you are doing stuff like that it's still important to look at relative worth. Ultimately 3% is still 3% even if it ends up being a few days. For example, if you can optimise your code in such a way to speed it up 5%, this will be greater than the 3%. Then again, if it takes you a few days to optimise your code in that way, perhaps even your 5% isn't worth it.

RAM isn't as cheap as it once was, it's still cheap compared to 5% but a 8GB stick, particularly a fancy CL9 one (which I agree with Vespine is rarely worth it) would be perhaps US$50 (presuming you don't sell the 4GB stick at a decent price).

It sounds like this is a bit late now, but US$50 could easily be the price difference between a i5 and i7 or a fancier i7 which will often give significantly more than 3%, particularly if it has a larger cache which it sounds like may be beneficial to you (which is actually another thing a lot of people benefit a lot less than they expect although in specific cases like yours where it's a matter of fitting or not fitting something specific it can make a fairly big difference).

It's unfortunately common people are sort of penny wise pound foolish or miss the forest from the trees with performance differences, paying a large amount for a small difference when they could get a far bigger improvement from the same or only a bit more.

From your description, you may very well get a bigger improvement than 3%, it's still IMO worth benchmarking for future consideration. Obviously not with 100 day calculations, but with shorter ones. Perhaps compare 16GB in different slots to give dual channel and single channel. (At least I presume this is still possible with Intel systems, I haven't actually looked in to but I think that's what the above review tried.) Then try the 28GB and 32GB (not a completely fair test but provided the memory usage isn't close to 28GB it won't be that bad).

Mind you DDR4 won't support multiple sticks per channel AFAIK, but still IMO worth testing.

Nil Einne (talk) 13:22, 14 August 2015 (UTC)

Thank you. I ordered three sticks of 8GB CL 11. The difference between that and a four-pack of the CL 9 stuff was on;y about $40. It was from Crucial, but Crucial does not list it as an upgrade for the new computer. That was one factor. Another factor is that Tom's Hardware Guide has data on the flex stuff, and on memory benchmarks, there is usually a big difference when the sticks are matched. The difference between 28GB and 32GB probably will not make a difference to me, but some of my things are so memory intensive that the memory speed from matched sticks is a significant factor. Bubba73 ^{You talkin' to me?} 16:10, 14 August 2015 (UTC)

Resolved

Bubba. I know this is marked as resolved, if you are still not happy with the performance, maybe these articles might inspire you Supercomputer, cluster computing, Next Unit of Computing. I'd love to have a reason to build a cluster computer out of Nucs. :) Vespine (talk) 23:16, 17 August 2015 (UTC)

lollipop op

does anyone know how to stop c phone from shutting down ,this only started after updateing lollipop o/s70.53.88.21 (talk) 16:48, 13 August 2015 (UTC)

I can't find any model phone which is commonly simply called a c phone. If you meant cellular phone, it's unlikely anyone can help you if you don't mentioned which model you're referring to. It's also unclear what you mean "shutting down". Does it shut down at random intervals even when the battery is still not empty? Does it shut down quickly after start up? Does it shut down when doing certain things? Nil Einne (talk) 23:16, 13 August 2015 (UTC)

Excel scrollbar

Hi,

I am working on an Excel spreadsheet which has an irritating feature - for some reason the horizontal scroll bar extends way beyond the area where there is actual data. I.e. the data range goes from Column A to Column FG but the scroll bar goes out all the way to column XES. This of course dilutes the scroll bar horrendously making it very difficult to move around the spreadsheet. I've tried deleting the excess columns in an effort to reset the scroll bar but this hasn't helped - Excel just recreates them. But there is no data in them...

Thanks, Peter — Preceding unsigned comment added by 168.252.233.190 (talk) 23:49, 13 August 2015 (UTC)

If you press CTRL-END, does it take you to the last cell of your data or some way off cell far away? Vespine (talk) 00:40, 14 August 2015 (UTC)

Takes it far away (original spreadsheet fixed but see below regarding another spreadsheet with enormous vertical scrolling). --168.252.96.5 (talk) 02:04, 17 August 2015 (UTC)

Also, if excel is really 'recreating' the cells, is there some macro or script that is in the spreadsheet that's using the cells in some way? Possibly a error in the code is causing the sheet to use up a bunch of cells... Vespine (talk) 00:41, 14 August 2015 (UTC)

No macros, .xlsx file (macros not enabled).

I've had that problem. Sometimes saving the file with a new name (Save as..) will fix it. Sjö (talk) 08:46, 14 August 2015 (UTC)

You mention deleting the columns - just to be clear, you need to highlight the columns, and then right-click and select 'delete'. Pressing the delete key clears the data from the cells but doesn't delete them as such. You also need to save the file afterwards. I've never had to save as a new file per Sjö's suggestion above, but you need to at least save it normally. 64.235.97.146 (talk) 15:38, 14 August 2015 (UTC)

Saving it normally after deleting them (and yes I was doing it properly) seemed to have fixed the problem. BUT I have another problem spreadsheet which has the same problem (although this time it scrolls vertically to row 1048576). When I try and delete these excess rows they just reappear as blank rows (but dilute the scroll bar). --168.252.96.5 (talk) 02:04, 17 August 2015 (UTC)

I once had a similar problem where the only solution I could find was to copy all of the data to a new blank spreadsheet. I can't explain why it happens. Dbfirs 12:27, 17 August 2015 (UTC)

The same kind of actions we used with the excess columns should work with the excess rows (delete and save). Assuming you're tried that, I would suggest clearing columns as well - that is, if you only have five columns of data, highlight columns F onwards and delete them as well. Excel has some strange ideas about what information to save regarding column and row areas and spacing and essentially 'clipping' the data you want to keep on both sides usually fixes it. If you're using a new enough version of Excel, make use of the .xlsx extension; that may also help. 64.235.97.146 (talk) 14:20, 17 August 2015 (UTC)