Jump to content

Talk:Binary prefix/Archive 12

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 5Archive 10Archive 11Archive 12Archive 13Archive 14Archive 15

Neutrality

This article isn’t the personal forum for advocates of the IEC prefixs (MiB, etc.) to get their digs in and try to slant the proper conveying of reality. I just fixed where the lead said “Nevertheless, the use of the SI prefixes as binary multipliers is still common in some areas” How about some objectivity and neutrality here? The use of MB and GB to denote binary quantities of computer RAM is a standard throughout the entire computer industry when it comes to how they communicate to their customer base. It’s used by all PC manufacturers in their advertising, brochures, literature, owners manuals, packaging, etc. And because of that, all periodicals—such as PC World—directed to a general-interest readership follow that convention. Further, the terms “MB” and “GB” are used by all—or virtually all—manufacturers and suppliers of aftermarket computer RAM. Does this whole article have to be slapped with a {neutrality} tag because editors with an agenda can’t behave themselves? I’ve corrected the above-cited slant of the facts to the much, much more accurate “Nevertheless, the use of the SI prefixes as binary multipliers is a ubiquitous industry practice.” Greg L (talk) 21:33, 11 August 2008 (UTC)

Likewise this is not the place for POV of the advocates of deprecating the Binary IEC prefixes. The use of use of MB and GB to denote decimal quantities of computer storage is a standard throughout the entire computer industry when it comes to how the computer companies communicate to their customers. In fact is is quite common to see both used on the same line of an advertisement as in "3 GB RAM and 500 GB HDD." In at least one case HP[1] clarifies 4GB SDRAM as 2x2088 but doesn't clarify the 320 GB HDD. A neutral position is that both are commonly used. So your fix needs to be undone. Tom94022 (talk) 21:53, 11 August 2008 (UTC)
  • What I wrote is that the use of the SI prefixes to denote binary values is a ubiquitous industry practice. That is an absolute fact. If you, Tom94022, can’t edit in good faith, there ware ways to get you to cool your jets. I suggest you go cool off on your own rather than edit disruptively here. Greg L (talk) 22:05, 11 August 2008 (UTC)
  • P.S.: What is now written is this:

Nevertheless, the use of the SI prefixes as binary multipliers when denoting the capacity of solid-state memory like random access memory (RAM) remains a ubiquitous industry practice: a standards group for the semiconductor device engineering industry, the JEDEC, defines "prefix to units of semiconductor storage capacity" using powers of two, clarifying in a note that the binary definitions are included "only to reflect common usage".

Whether you like “reality” or not, has no bearing of whether “reality” should be accurately conveyed. The above has the virtue of being 100% true and is an accurate and balanced reflection of the current situation. Only balanced truth is truth. Let’s keep it that way, shall we? I’m quite done for the moment with all this horseplay. I gotta go. (*iTunes*) Greg L (talk) 22:18, 11 August 2008 (UTC)
Last time I checked Flash memory was a solid state memory but when packaged as a USB Drive its capacity is marked using decimal prefixes. So the semiconductor industry usage is not ubiquitous and thus your edit not 100% true. That's why I preferred:

Nevertheless, the use of the SI prefixes as binary multipliers is still common in some areas: ...

which I believe to be 100% true, fair and balanced. You can fix the statement by limiting the industry's scope but as it stands I don't think it rises to the level of accuracy and balance appropriate for Wikipedia Tom94022 (talk) 23:08, 11 August 2008 (UTC)
  • Jeez. Track the logic of the statement. “…transistorized memory”: what does that include? Well, L1, L2, and L3 caches; and (notably) RAM. And what does it exclude? Transistorized mass storage like thumb drives, which must be formatted like a spinning hard drive. Please try to get it through your head. And the wording you’re complaining about is less encompassing than that in the second half of that same sentence, which quotes JEDEC: “semiconductor storage capacity.”

    Don’t be pain please. The slant that was in this poor article before went beyond all reason. Wikipedia is not a tool for you to use to advocate change in the way the world ought to work by slyly backhanding reality. The prior wording was crafted to imply that there were odd holdouts in the computing world that still embrace a long-discredited practice. While that certainly may be what you believe, it is far from an accurate portrayal of the truth and it was intellectually dishonest for the author (whomever that was) to have written that in the first place. The use of the SI prefixes to denote the capacity of transistorized memory is a widespread, ubiquitous, standard practice within the computing industry and is also universally observed by the all general-interest computing magazines; you don’t see “2 GiB” of RAM in PC World today and it’s doubtful anyone will see it ten years from now. That’s the simple reality. Deal with it. Greg L (talk) 01:47, 12 August 2008 (UTC)

Last time I checked Flash memory was transistorized memory, not that it matters at all to this debate. BTW, why are u injecting this new term into the debate instead of acknowledging that your edit is incorrect? In the real world the terms GB and MB are probably used at least as often with decimal meaning as they are used with binary meaning, perhaps more often especially now with the ubiquitous presence of flash drives. That is the real world that you simply refuse to acknowledge - perhaps u should try to deal with it and stop trying to put your POV into every article. Tom94022 (talk) 05:37, 12 August 2008 (UTC)
The bit above you quoted at 23:08, 11 August 2008 presents a biased point of view because it implies deprecated use which is contrary to the real world situation. I again note you have not answered the challenge yet. Fnagaton 08:11, 12 August 2008 (UTC)
  • Tom94022, regarding your above 05:37, 12 August 2008 post where you write “In the real world the terms GB and MB are probably used at least as often with decimal meaning as they are used with binary meaning, perhaps more often…”, you are failing to understand the point of the first three paragraphs as well as the sentence I modified.

    The preceding paragraphs of the article establish that the new IEC proposal provides new binary prefixes and that is intended to liberate the SI prefixes to exclusively denote their proper decimal meaning. The sentence I changed (I didn’t add it) establishes that notwithstanding the availability of the new IEC prefixes, the common use of “GB” and “MB” to denote the binary meanings continues to be a ubiquitous industry practice. Again, I didn’t add that sentence. But it does need to be there in order to accurately complete the thought.

    Further, the sentence can’t be slanted by stating something that implies that the IEC prefixes are in wide use nowadays and are only used by a few dillweed places (“Nevertheless, the use of the SI prefixes as binary multipliers is still common in some areas”); other than for some internal documents by the standard bodies themselves (whom the average Joe has never even heard of), the entire God-damned real world is completely ignoring the IEC when communicating to common folk.

    The simple facts can be boiled down to this:

  1. The metric prefixes are used for both the binary and decimal values in computing
  2. When splitting hairs, the distinction sometimes matters
  3. In an attempt to remedy this confusion, the IEC proposes exclusively-binary prefixes like “kibibyte” (KiB) that make people sound like a Klingon or Porky the Pig
  4. The computing world soundly ignores the IEC. (*sound of a phonograph needle being ripped off of an LP*) As a consequence, someone would get laughed clean out of the computing store if they waltzed in all fat, dumb, and happy and announced that they wanted “a PC with three gibibytes of RAM ‘cause I read on Wikipedia how terminology like that is the new emerging standard and the old-school way is only used in some quarters and I wanted to sound really smart.”
  5. After three years of an ill-thought-out experiment trying to make Wikipedia promote the use of the IEC prefixes and after having that practice abandoned, Thunderbird2 and Tom94022 continue to try to make it sound like the IEC proposal has been widely adopted and there are only some odd islands of “old school” where the old practice continues
  6. Fnagaton, Greg L, Swtpc6800, and Woodstone try to edit Wikipedia to help make it more factual
  7. The whole experience of trying to correct the article proves to be about as much fun as a colonoscopy because of Tom94022’s problem with WP:POINT. In the mean time, Thunderbird2 gets blocked for 24 hours for being intransigent and editing furiously against the rules and making it double-hard to get anything done here.
Now… you know all this. So stop badgering us and being a pain. Greg L (talk) 14:51, 12 August 2008 (UTC)

Lawsuits

I deleted the lawsuit-related sections. This sort of content (“…a lawsuit was filed in the Superior Court for the City and County of San Francisco…”) is a bit of a grey area but this stepped over the line and is inappropriate on Wikipedia. Employing such an editorial tool—citing lawsuits to illustrate the shortcomings in products and practices—should only be used when it is especially notable and germane to the subject, such as when there is corporate malfeasance and serious harm to public safety (asbestos or cigarettes, Thalidomide). Citing lawsuits is not a proper tool to help editors promote a particular point of view in taking sides on an issue of standards in units of measure. What would Wikipedia look like if editors started trying to demonstrate the safety shortcomings of lawn mowers by adding “lawsuit” sections to articles? Lawsuits—particularly American lawsuits—could be used to impeach absolutely anything. Citing lawsuits is not an encyclopedic practice and in this case seemed an agenda-based effort by editors trying to demonstrate that the computing industry has been/currently-is doing the wrong thing by not rapidly adopting the IEC prefixes. Wikipedia exists to convey facts in an encyclopedic manner and is not to be used as a tool to promote change in how the world works.

P.S. Of the three products I cited above (asbestos or cigarettes, Thalidomide), only one of those articles (asbestos) mentioned lawsuits. We don’t need this sort of stuff to show how without “mebibyte”, the term “megabyte” has shortcomings—that much is clear enough from the rest of the article, which makes that point clear in a factual and mathematical way. Greg L (talk) 15:50, 13 August 2008 (UTC)

Wikipedia should warn people that their new BMWs may have had scratches in the paint. The court case made it to the Supreme Court of the United States. BMW of North America, Inc. v. Gore -- SWTPC6800 (talk) 18:08, 13 August 2008 (UTC)
As someone who supports having different prefixes for the binary and decimal units, I'm certainly happy to see the lawsuits go (they were against manufacturers who used the prefixes in their "correct" sense after all), but it can be argued that the lawsuits are of relevant interest to the article, at least in the history section. shreevatsa (talk) 02:47, 14 August 2008 (UTC)

History

The history section actually doesn't mention the IEC units :) I get the impression that it was the NIST which came up with the kibi- etc. units (what a terrible choice of names, really), and the IEC adopted them "with significant input from the [NIST]", as the NIST press release says. I don't know how to research this, could someone insert the correct facts? (E.g.: when did IEC adopt them, when did IEEE etc.) shreevatsa (talk) 03:43, 14 August 2008 (UTC)

Reference number 1, the NIST's "Prefixes for binary multiples" is in part based on Bruce Barrow's "A Lesson in Megabytes". This is a short essay written for the IEEE Standard's monthly newsletter. A copy of the January 1997 newsletter can be found here: [1] -- SWTPC6800 (talk) 01:18, 21 August 2008 (UTC)

Software

The Binary_prefix#Software section list 20 minor utility programs but does not have any software a major publisher. Nothing from Adobe, Apple, Microsoft, Novel, Oracle, Red Hat or Symantec. Some of the packages listed are from a one or two person publisher. From a reliable source perspective, this is using nothing but personal blogs while omitting all major newspapers and magazines. A search of the Red Hat site produced this: 'Your search - mebibyte - did not match any documents. No pages were found containing "mebibyte".' The whole software section could be replaced by a sentence saying virtually all software companies ignore the IEC binary prefixes. -- SWTPC6800 (talk) 01:18, 21 August 2008 (UTC)

It is certainly important to point out that most software uses the same prefixes that have been in use for hundreds of years. The section begins with "As of 2008, most software uses the traditional binary units." Maybe that sentence is worth expanding on, but I don't know how. At the same time, it is also important to point out cases of actual adoption, so the list of examples is appropriate. The list also gives a good feel for the kind of software that has adopted the new prefixes -- if it contains only "small fry" software, then the reader's impression is that those are the software that uses the new prefixes. (Although I would argue that the Linux kernel, Launchpad, Pidgin etc. are indeed important examples.) shreevatsa (talk) 02:57, 21 August 2008 (UTC)

[Unrelated to the above] I notice that Apple Mac OS X's 'df' utility shows either "Gi"/"Mi"/"Ki" etc. with 'df -h', or 'G'/'M'/'K' etc. in the decimal sense with 'df -H', so there is some example of adoption by Apple. This is not reflected in the manpage, so it's not clear if this is something that can be cited. (Their implementation seems to be to simply add an "i" when using binary units, so they show "0Bi" for 0 bytes :D) shreevatsa (talk) 02:57, 21 August 2008 (UTC)

That sounds like a typical IEC binary prefix application. An undocumented command line option on a program that virtually no Mac user knows about. -- SWTPC6800 (talk) 03:59, 23 August 2008 (UTC)
:) I don't know if your "virtually no one knows about" refers to the option or the program. I didn't go looking for it; I just noticed it (after ls and cd, du and df are probably my most used commands these days; I'm running low on disk space.) It's a pretty bad implementation though, done without much thought or discussion or documentation or correctness. Shreevatsa (talk) 06:27, 23 August 2008 (UTC)

Terminology

This issue has cropped up before (search archives) but it has never been fully resolved, so bringing it up again. There are two sets of prefixes here, going by "mention": the kilo/mega/giga set, and the kibi/mebi/gibi set. And there are at least three sets of prefixes, going by "use": the kilo/mega/giga set used in the decimal sense, the kilo/mega/giga set used in the binary sense, and the kibi/mebi/gibi set used in the binary sense. (There are also others, like the kilo/mega/giga set used in the approximation sense, where 32K+32K=65K, but probably less important.) So we need to decide on names for all of these, and be consistent throughout the article. It currently uses "SI prefixes" for the first of the three, "traditional binary prefixes" for the second, and "IEC prefixes" for the third. (The latter two are somewhat of a misnomer, because "traditional" depends on what area (and concomitant tradition) one is speaking about, and "IEC" does not reflect the fact that the IEC was neither the first to come up with the prefixes nor the only standard to have embraced them.)

It would be good if someone came up with better names. At any rate, though, calling the second one "common practice prefixes" is misleading, because it's the common practice in some areas (computer memory) but not in others (speeds, drives). As someone says

"Everyone knows that 1MB is 1024KB, unless you’re talking about DVDs, or reading manufacturer specs for a hard drive, and that’s just the hard drive manufacturers being stupid. Everyone knows that ‘K’ on a computer means 1024; except for speeds, where it means 1000, except for file download speeds where it means 1024, except when it’s the speed of your modem, when it’s 1000. Everyone knows that. What, are you stupid?"

Shreevatsa (talk) 14:01, 21 August 2008 (UTC)

Agree with the analysis above, but actually the article has two names for the binary use of K/M/G: it uses the word "traditional", but it is also full of phrases like "common use", "commonly applied". So I had aligned the article with "common practice". Using the word "traditional" is misleading for such a young practice of alternative use of a much older prefix. I'm open for a better term, but we should try to keep to only one for each set and use it consistently. −Woodstone (talk) 15:59, 21 August 2008 (UTC)
I agree, we should strive for consistency. As far as I can tell, the word "common" only occurs in the article when describing usage (to say that "traditional binary prefixes" (or whatever) are actually commonly used in various settings); the units/prefixes themselves aren't actually called "common units". The term "traditional binary prefix" is good because (i) the names themselves are traditional (kilo etc. as opposed to the newfangled kibi etc.) and (ii) for binary prefixes, they are what have been traditional. On the other hand it is bad and misleading because (as you say) it had definitely not been traditional to use them in the binary sense; they have been, and are still used, chiefly in the decimal sense ("one kilo of sugar"). But calling them "common usage prefixes" would be just as problematic, because (again) they are (currently) common usage in computer memory but not common practice in speeds or drives (or, more generally, distances, weights...)
The article previously called them "SI prefixes used in the binary sense", but some people thought it was biased, others thought it was inaccurate, and I thought it was too much of a mouthful, anyway. In a sense, the absence of a proper name for them is a reflection of the state of things: they were never officially decided-on and named, but just came to be. This makes it hard to give them a name... Shreevatsa (talk) 16:36, 21 August 2008 (UTC)
I think the problem stems from how you started this section: two sets of prefixes for three sets of usage. Apart from the kibi set, there no binary prefixes, there are only prefixes used in binary sense. Perhaps we could stick to adjectival description:
  • binary prefixes kibi, mebi, gibi
  • SI prefixes kilo, mega, giga
  • binary intended prefixes
Woodstone (talk) 17:35, 21 August 2008 (UTC)
Yes, that's what the problem is. If there is consensus that "binary intended prefixes" is an appropriate name, go ahead and change it. It seems somewhat awkward, but so is this whole issue, anyway.... For the first set, "IEC prefixes" is the name currently used, but not much outside Wikipedia. "Binary prefix", in the broader sense of this article, just means any prefixes used at any time for a binary purpose. Shreevatsa (talk) 05:02, 23 August 2008 (UTC)
How about "ad hoc binary prefixes" or "de facto binary prefixes"? Tom94022 (talk) 17:09, 25 August 2008 (UTC)
The problem is that they are not "binary prefixes", since most of the time (and offically) their meaning is decimal. The are at most "prefixes used the binary way". −Woodstone (talk) 18:03, 25 August 2008 (UTC)

"files" section

The examples given in here need to either be rewritten or removed. The following problems are apparent:

  1. The "GNU/Linux" and "Solaris" sections are muddy, overly-technical and unclear. "Command line and file manager" are not programs under these operating systems; the "ls" command and the "Nautilus" file manager are. Because both examples use the same programs, there is no reason to use both. We should present this as "Unix-like systems" and name the specific programs used. This also avoids triggering the GNU/Linux naming controversy by having to pick between using "Linux" and "GNU/Linux".
  2. The "Mac OS X" entry is unsourced. Mac OS X uses a Unixish "ls", so should return the same result as the other Unixes for command line results. The Mac OS X Finder should be given a verified answer.
  3. No secondary sources. We need secondary, not primary, sources here to ensure that we're not doing original research.

Given these problems, I feel that the examples should be removed pending a rewrite. They'll be available in the page history for anyone who wishes to try. Chris Cunningham (not at work) - talk 11:59, 24 August 2008 (UTC)

I agree; that section is poor as it stands. Shreevatsa (talk) 15:00, 24 August 2008 (UTC)
I agree the section is poorly written but I don't think that removing it is an appropriate action, why don't u just rewrite it? BTW, the sources used are acceptable under WP:OR. Tom94022 (talk) 22:35, 24 August 2008 (UTC)
I dislike "I spy" games on Wikipedia, where arguments are advanced using editor investigation using primary sources. If I find time I will re-add the section in a more appropriate form. However, in its present state it was superfluous when considering the prose above it. Chris Cunningham (not at work) - talk 08:58, 25 August 2008 (UTC)
It really wasn't superfluous since it clarified which UNIXs did what. Perhaps it could have be moved to footnotes but removing it removes valuable information. Also, the sources cited are either secondary sources (the software operating on a system being the primary source) or in the alternative acceptable primary sources. Either way, IMO not an "I spy" game Tom94022 (talk) 17:07, 25 August 2008 (UTC)
This is not a random collection of useful information. The article is long and rambling and not getting better fast. It has to get shorter if it is to be readable. Losing the anecdotal factoids is a good start. Chris Cunningham (not at work) - talk 18:22, 25 August 2008 (UTC)
Given Chris Cunningham's comments above he should have removed the section detailing the software that claims to use binary prefixes because those primary sources are even more dubious than the section he removed. Fnagaton 07:00, 25 August 2008 (UTC)
There's a good argument for that as well. However, that section does not have the problems with vagueness and disputed terminology that the OS examples did. Chris Cunningham (not at work) - talk 08:58, 25 August 2008 (UTC)
U actually made the article more vague by removing the clarifications regarding the UNIX family. Now what? How long before your rewrite? Is it a good idea to remove something, essentially correct but poorly written, pending an uncertain update? Why don't we restore the information and then u can change it at your leisure? If you won't, would u object if I restored it as a series of footnotes? Tom94022 (talk) 17:07, 25 August 2008 (UTC)
Yes, I'd object. It's no less worthwhile as a set of footnotes, which are also meant to be backed up by secondary sources. This article needs a serious rethink if it to improve to the level of a good article, and that's not going to happen while someone strenuously objects to the removal of any part of it, even on a temporary basis. Anyway, this information probably belongs on Timeline of binary prefixes, the sub-article which I remember took us so long to agree shouldn't live on this talk page forever. Chris Cunningham (not at work) - talk 18:22, 25 August 2008 (UTC)
Fine you object but will u start a revert war if i but back the Linux and Solaris citations as footnotes? The manuals cited are reliable secondary sources or at worst acceptable primary sources. Tom94022 (talk) 06:13, 26 August 2008 (UTC)
Brinkmanship isn't particularly constructive. I've already indicated that the article needs to get shorter. I've rewritten the section to remove the unsourced claims already present and to allow expansion. Please try to keep it to the point instead of being endlessly inclusive. Chris Cunningham (not at work) - talk 07:55, 26 August 2008 (UTC)
You are the one who pulled the cites notwithstanding my objection; I'm trying to work it out in talk and now u accuse me of "brinksmanship". Yr latest edit may be incorrect, at least as far as one Unix, True64 Unix (HP/Digital) which does not appear to use prefixes as of 2001.[2] You did remove two sourced facts - on what basis? On the whole, in my opinion, yr edits have made the section shorter but less factual. Since you seem to feel free to go ahead and make changes not withstanding discussion, I will now attempt to edit the section back to something better than where u left it. Tom94022 (talk) 15:55, 26 August 2008 (UTC)
Being continually told to revert because primary sources are fine is not "discussion". As for the Tru64 thing, the addition of the word "most" has covered this for now. Chris Cunningham (not at work) - talk 21:29, 26 August 2008 (UTC)
Being continually ignored about the issue as to whether the references are secondary sources or acceptable primary sources is also not "discussion". I have been thru this once before and the consensus was that software reference manuals are reliable sources usable in Wiki articles. Yr edit "most" is entirely unsourced; you continue to make the section worse. When I have a few more moments I intend to correct it and we shall see what happens from there. Tom94022 (talk) 23:32, 26 August 2008 (UTC)
As a participant in the previous discussion, I believe the "consensus" Tom94022 refers to is here and here.
Cheers,  This flag once was red  11:38, 27 August 2008 (UTC)
Yeah, I was there too. I'm really discussing this for the benefit of any onlookers, not Tom94022. Chris Cunningham (not at work) - talk 12:09, 27 August 2008 (UTC)

Since we are in the mode of educating onlookers, here is the end of the discussion at the OR noticeboard:

  • There is no rule against using primary sources. We certainly can cite a manual to confirm that an OS displays memory/disk values in a certain way if that is stated in the manual. *** Crotalus *** 18:51, 3 March 2008 (UTC)
o Crotalus is correct... caution should be used in discussing primary sources... but there is no rule against using them. In fact the policy specifically states that you can use them: "Primary sources that have been published by a reliable source may be used in Wikipedia..." Blueboar (talk) 19:25, 3 March 2008 (UTC)
  • One problem I found was that some of the references took the form "I did this and ran this command; here's the output". This seems to me to be original research. On the other hand, I think reference a manual (i.e. man page) and indicating something like "ls outputs files using 1024 blocks when such and such a switch is used" is fine, since the manual says as much. The problem with screenshots is that they would be displaying output for a particular file; you would need to use inductive reasoning to argue the general case (i.e. that the file properties of a particular file is representative of all files). As such, I'm not convinced this could be considered a legitimate source. Andareed (talk) 07:06, 5 March 2008 (UTC)
  • It was me who objected to the man page; I still have some reservations (the page referenced appears to contradict the claim that Linux uses 1024 instead of 1000: "SIZE may be (or may be an integer optionally followed by) one of following: kB 1000, K 1024...", it's less than clear to a layperson, and is less than authorative - though this last argument could be dispensed with by changing the source to, say, GNU) but I'm prepared to accept any consensus. I agree with Andareed that the main concern was that the "references" consisted of screenshots and "speak to your friendly neighbourhood computer whizz, he'll totally agree with what I'm saying, man." I'm also slightly concerned that the arguments of 5 editors are being misrepresented as an opposition to primary sources; my opposition was to the interpretation of those primary sources and the refusal to provide secondary sources supporting those interpretations. Finally, it's worth noting the comment beside the Apple listing: "informed guess because of its Unix ancestry and comments elsewhere, explicit version and similar verification required because unavailable to the author at this time".

 This flag once was red 

So why do you persist in eliminating valid source material? FWIW, I agree that the Apple comment (while correct) should be referenced, but that is a reason to tag it, not to eliminate it. The section as it now stands is less correct than when u began your editing. Tom94022 (talk) 15:59, 27 August 2008 (UTC)