Wikipedia:Reference desk/Archives/Computing/2013 May 21

= May 21 =

computer stuff :3
I just wanna know if its possible to make it so when i turn on my computer and start up the pc profile. If so, how? — Preceding unsigned comment added by 173.89.163.134 (talk) 01:27, 21 May 2013 (UTC)


 * What kind of computer do you have? Are you referring by any chance to Boot Camp on a Macintosh computer?  Tarcil (talk) 01:51, 21 May 2013 (UTC)


 * Sounds like it. With Boot Camp, you can run "PC" OSes on a Mac. --Yellow1996 (talk) 01:20, 22 May 2013 (UTC)

What is the threshold of date/time difference for certificate error?
When I was trying to use https connection with a wrong system time I would receive a notification of certificate error. How much difference in system time and the certificate time is needed for the system to think that the certificate is not valid，e.g. 1 minute late? 1 days earlier?--朝鲜的轮子 (talk) 09:34, 21 May 2013 (UTC)
 * This depends on how the certificate is validated. The Kerberos system permits a default clock skew of 300 seconds (5 minutes) but is configurable by the system-administrator.  Your https certificate may be subjected to any number of distinct client- and server-side technologies for validation.  For example, I often use mod_auth_pam and this way my HTTPS connection uses all my ordinary pamd settings.  Commercial web servers may use apache and pamd, or Microsoft Windows Server, or any number of alternatives.  Nimur (talk) 13:43, 21 May 2013 (UTC)

Google Webmaster tools - Search results URL removal tool
If someone removes a specific URL, for how long will Google automatically ignore it?, and in other words, if it was removes, is there any period of time that in which the URL will automatically be available in the search results again? (for example, when the webmaster unknowingly open a new pages with the same alias\Url?. Thanks. Ben-Natan (talk) 12:08, 21 May 2013 (UTC)


 * Well, according to this page (straight from Google), once you remove a url it will automatically be off the Google Search results. --Yellow1996 (talk) 01:07, 24 May 2013 (UTC)

project based question
sir i wish to develop a software in 3D virtual technology.because in new generation the it developed is so far.. we chat with one other person with video or voice likely. what i mean i wish to develop a software when user stand in one side and talk the capture element may be a camp or phone camera. in which it can select the full body of the specified element and project it in to the other direction (opposite side of the user). were he can see the full motion of the other one.

so what i mean please give me some guidance about this question and help me out of this.its my dream project.and give some basic about it. means specified sites ,professors etc.... — Preceding unsigned comment added by Anila g pillai (talk • contribs) 16:25, 21 May 2013 (UTC)


 * You might start at holography. ¦ Reisio (talk) 16:37, 21 May 2013 (UTC)

Graphics slow on ipad
Hi all, I'm doing a simple ios app (so simple it embarrasses me) and can't get it to work sensibly. The app involves a graph of about 10000 bezier path points, but these take forever to draw on ipad. It is reasonably quick on the simulator, but somehow tediously slow on the device. It takes about 3 seconds on simulator, and about 50 on the ipad. I'm getting the feeling the problem may come down to something other than the drawing itself, even though (having studied the log file) it is definitely the drawing line ([path stroke]) that takes up the time. So the fault is with that line of code, but it may be not so much the drawing, but some calculation that takes place along with the drawing. The reason for my deduction is that an ipad will run slower than the simulator, except for rendering graphics, where the graphics accelerator will come into its own. Since the ipad is a lot slower than the simulator, I deduce that it is not the drawing bit, because that should not be so much slower.

So my guess is that it may be something like doing rounded corners with the "kinks" in the graph, that is, with the app trying to draw pretty corners whenever the graph zigzags. If it is getting hung up about making them look good, it might sit there calculating every corner precisely. Does anyone know how to speed this up with really simple graphics? Thanks, IBE (talk) 20:14, 21 May 2013 (UTC)


 * Seems like you answered your own question: simplify the graphic. Another more drastic approach I’ve seen before is to query a remote server which returns a raster image. ¦ Reisio (talk) 20:57, 21 May 2013 (UTC)


 * Thanks - but is there a way of just simplifying the drawing process itself? When I draw the pattern under the graph, without drawing the graph, it actually draws quickly. The resulting image is a blue background on top, and a criss-cross pattern below, but the pattern is clipped to the zigzag lines of the graph. It's the same graph without the line to make it snappy. It isn't presentable without the line, but I don't need anything really slick. I don't know what settings or drawing classes avoid this funny business (and technically, I'm not sure it's the computational aspect of drawing, I'm just surmising). If anyone knows, I'd be grateful, because I'm sure this shouldn't happen - it's the most pathetically simple thing and somehow it isn't happening. IBE (talk) 22:37, 21 May 2013 (UTC)
 * I don't know what's involved but that seems very slow indeed. You might be able to make one big call instead of 10000 to do the drawing if its just a sequence of bezier curves. Dmcq (talk) 23:27, 21 May 2013 (UTC)


 * There is no substitute for profiling. One might speculate that it's due to floating point performance, but the iPad has a pretty decent fpu. One might also speculate that your strategy for double buffering is the problem - see this discussion for some insights into that. -- Finlay McWalterჷTalk 08:24, 22 May 2013 (UTC)


 * My suggestion is to take a different approach: why are you drawing 10,000 points? Why not refactor the process so you only need to draw 10 or 100?  Can the user meaningfully see and interact with 10,000 distinct entities?  If you improve performance in an O(n) fashion, you still have not produced a scalable algorithm.  This is critical: even if you tune performance on today's iPad with 10,000 points, are you sure it will work on next year's gadget, or next year's data set with 100,000 points?
 * A better approach: why not draw only ten or one hundred points at any given time? If the user needs higher resolution or more detail, provide a user-interface (say, zooming in) to render in higher detail on a subset of the data.  This is how GIS maps work: it's intractable to draw every side-street in the world at the same time; at any given zoom-level, reasonable map software is only drawing a few tens or hundreds of roads at a time.  The software uses a smart data structure - say, a spatial index database or an R-tree - and an appropriate algorithm to select the subset of data that is relevant for any particular instant.   Nimur (talk) 11:48, 22 May 2013 (UTC)
 * It is so slow I wonder if there is some interaction problem between the curves or between the curves and something else. If each bezier curve had a check with every other bezier curve for instance the time after every 1000 would go up as a square. Or from what I read above are the bezier curves forming the edge of a mask so everything drawn before is checked against them - how many things are drawn before or is the check done every time a curve is drawn or only at the very end when the mask is complete? Dmcq (talk) 12:14, 22 May 2013 (UTC)

Thankyou all ..... I won't know what the problem is for a while, but this is a great set of ideas and resources to start with. I was going mad, so this gives me a way forward. I agree with the suggestion that it may be an interaction problem - i can see it is rather weird, and the solution, when i find it, is bound to be very simple. IBE (talk) 13:10, 23 May 2013 (UTC)