Wikipedia:Reference desk/Archives/Computing/2015 August 18

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 18

Contribution to page Lanczos_resampling

Hi All,

I wanted to contribute to the above page, because the interpolation formula gave me inaccurate results, and I have found a more precise one. However, my editing was identified as "potentially unconstructive" and was denied. I have tried to report it as false positive, but the same result there. Below is the content I wanted to add to the false positives page.

Bets regards: Balázs Bámer

Wikipedia is an encyclopedia. Our task is to document the world as it is - not as we wish it to be. The article you were editing is there to describe how this method works. Your idea for an improvement has nothing whatever to do with how the method currently works. That's why your addition to the article - no matter how clever, inventive or useful - was completely inappropriate. That's why it was subsequently (and 100% correctly) deleted.

FWIW, your more precise version probably takes more CPU cycles to compute - and that really matters with image filtering algorithms. So I suspect that, while you may make some improvement on precision, it's not enough to justify a slower execution time in applications where this approach is most commonly used.

SteveBaker (talk) 16:49, 20 August 2015 (UTC)

Bamerbalazs

Username

Bamerbalazs (talk · contribs) (filter log)

Page you were editing

Lanczos_resampling (filter log)

Description

I was trying to add the following lines to part "Interpolation formula". Reason of editing is that I have used the formula in the original description and it gave me inaccurate results. My modification solved this problem.

However, due to the finite precision of calculations the above form may introduce oscillations in the interpolated signal. Thus it is better to perform the calculations this way:

$N(x)=\sum _{i=\lfloor x\rfloor -a+1}^{\lfloor x\rfloor +a}L(x-i)$

$S(x)=1/N(x)\sum _{i=\lfloor x\rfloor -a+1}^{\lfloor x\rfloor +a}s_{i}L(x-i)$

Date and time

07:06, 18 August 2015 (UTC)

Comments

Any change to a Wikipedia article must be based on reputable published sources. If your formula has not been published, we can't use it, regardless of how good it is. This is a fundamental Wikipedia policy. Looie496 (talk) 12:49, 18 August 2015 (UTC)

Thank you. I don't think i have a chance to publish it - although it must be already known as this is a very common numeric problem. Bamerbalazs —Preceding undated comment added 16:19, 19 August 2015 (UTC)

Should I accept Microsoft's suggestion of upgrading my Windows 7 by Windows 10, for free, after they found my computer to be appropriate.

77.127.249.96 (talk) 09:21, 18 August 2015 (UTC)

The proper questions is: Should I ignore the obviously intelligent choice of upgrading to a better operating system and keep my older and less supported operating system? 209.149.114.69 (talk) 12:10, 18 August 2015 (UTC)

To which the answer might well be "Yes" if its usability is greater and there are no features of Windows 10 which the OP actually needs. Many PCs continue to run quite satisfactorily on Windows XP. The problem is, we don't know what the OP does with their computer, so can't really guess any form of answer to their question.--Phil Holmes (talk) 12:23, 18 August 2015 (UTC)

"Many PCs continue to run quite satisfactorily on Windows XP." -- That is correct if you are a spammer or a botnet operator. If you are the actual owner of the computer, it is not entirely correct. I don't even try to remove junk from XP computers anymore. I just fdisk/format/reinstall. 209.149.114.69 (talk) 13:21, 18 August 2015 (UTC)

I run a small netowrk of mostly XP computers with no assistance whatever to either spammers or botnets. No junk on the PCs in my control.--Phil Holmes (talk) 15:46, 19 August 2015 (UTC)

People still fail to understand that the "junk" in their computers is not the fault of the particular OS but the fault of the person(s) using this OS.--Lüboslóv Yęzýkin (talk) 01:17, 21 August 2015 (UTC)

That is of course a ludicrous question. Why on Earth would someone upgrade straight away before all the inevitable compatibility problems and security holes have been patched? The OP should do the intelligent thing and wait at least several months, as I am doing. Pretty much everybody I know who has upgraded has had some kind of issues with this release. 131.251.254.154 (talk) 13:40, 18 August 2015 (UTC)

You have a year to decide, Microsoft is offering this to you for free for the first year of release, so I think July 29th 2016? Don't rush into it if you don't feel you'll need it but hey, free is free right? 81.138.15.171 (talk) 12:42, 18 August 2015 (UTC)

Well, how docile do you want to be? If your computer is "appropriate" for Windows 10, it will also be "appropriate" for something quite different. There are many options available. (I happen to be using Mint LMDE right now. It works well for me.) -- Hoary (talk) 13:48, 18 August 2015 (UTC)

You might want to wait a couple of months and then do it, unless you have a pressing need. In the first two days of Windows 10, I upgraded three Windows 7 computers and one with Windows 8.1. It had no major effect on two of the Windows 7 computers. The other Windows 7 computer had been very sluggish, and now my daughter says that it is no longer sluggish, so that is a big improvement. The W8.1 is another story. It came with W7 and I upgraded to 8.0 and then to 8.1, but I'd been having problems. For the last few months, a few pieces of software just would not run properly and the software makers could not figure out why. I hoped that W10 would fix the problems. It did fix one problem, but not others. Then 8-10 hours after the W10 install, it started saying that the network card was disabled. Rebooting fixed the problem. It happened once more that day - reboot fixed it. It started happening more and more the next day, about 10-12 times, and finally a reboot would not fix it. I tried everything I could think of. I took it to be repaired and they tried everything they could think of, couldn't fix it. They think it has something to do with W10. So, that computer was my main computer, so I ordered a new computer with W10 installed, to arrive tomorrow. Bubba73 ^{You talkin' to me?} 19:52, 18 August 2015 (UTC)

Yes, as the saying goes, "the plural of Anecdote is not Data". We don't have data on how many suffered and how many benefited. For my originally Win 8.0 machine I was very pleased with the 8.1 update, and also pleased with 10 which gave me no trouble. Many of my fellow upgraders of course will have a similar anecdote, and many others will differ. How many? We don't have data. Jim.henderson (talk) 20:03, 18 August 2015 (UTC)

I haven't heard that saying before - I like it. I'm going to send it to a friend (we are always arguing aout the value of anecdotal evidence.) Bubba73 ^{You talkin' to me?} 20:14, 18 August 2015 (UTC)

The only issues I've had were that I had to manually update my video card drivers to get anything to work, and had to give Microsoft my billing information so they could charge me $0.00 to redeem my free copy of the Windows 10 Minecraft beta. Other than that, they appear to have done their job in beta testing. I only hear about people having issues from people who have not upgraded, not from anyone who actually has upgraded. This OS is to Windows 8 what Windows 7 was to Vista. Ian.thomson (talk) 01:30, 21 August 2015 (UTC)

The main criticism I've heard about Win10 is that it phones home, unless you take specific and difficult-to-discover steps to avoid it, and perhaps even then. Users generally don't find out about phoning home, so you probably wouldn't hear about it. --Trovatore (talk) 20:27, 22 August 2015 (UTC)

Do TCP & UDP function the same way over IPv6 the same way they do over IPv4?

Do TCP & UDP function the same way over IPv6 the same way they do over IPv4? If there are changes, what are they? — Preceding unsigned comment added by 134.242.92.2 (talk) 14:14, 18 August 2015 (UTC)

The addressing scheme changes. Nothing else does for TCP/UDP. TCP/UDP are a different layer than IP. Implementation will change, due to the addressing scheme change. For example, if I hard-coded a 15-character string for all addresses in my implementation, I'd have to change it. 209.149.114.69 (talk) 14:56, 18 August 2015 (UTC)

The checksum computation changes, as described in RFC 2460, section 8. -- BenRG (talk) 20:17, 18 August 2015 (UTC)

There's also RFC2147 TCP and UDP over IPv6 Jumbograms. When you buy IPv6 network hardware, you can ask the vendor about support for such extensions.

In practice, a lot of IPv6 traffic is actually a software layer on top of TCP on top of IPv4 networking hardware - this is called 6to4 and it has been commonplace across the industry because it's cross-compatibile with older networking hardware - see, e.g. Routing IPv6 over IPv4 (2000); so if you put TCP on IPv6, you'll probably get software TCP on top of software IPv6 on top of hardware TCP on top of hardware IPv4 on top of a hardware network PHY.

If you want more information, Cisco publishes a book: IPv6 Implementation Guide.

Nimur (talk) 15:43, 18 August 2015 (UTC)