Jump to content

Template talk:Quantities of bytes

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

-ibi-bollocks

[edit]

@people reverting the edits to this: The list in its unedited state effectively doesn't contain the most widely used 1KB=1024B notation. The IEC poppycock is so rarely used that even including them in a table like this is just trying to push them down people's throats. If you really think it's necessary to even include those ridiculous -ibi units in this table, don't make them the primary/only notation in the table. Boatmurdered (talk) 00:57, 6 February 2012 (UTC)[reply]

They're not primary/only, and we include them for completeness and lack of ambiguity. --Cybercobra (talk) 01:20, 6 February 2012 (UTC)[reply]
The table has been reverted to a state where the binary value column is duplicated, and there is no way to quickly know that the binary values are slightly larger than the decimal values. In both cases, I am having trouble seeing this as anything but a step in the wrong direction.  :-( —Quantling (talk | contribs) 15:26, 6 February 2012 (UTC)[reply]
There is duplication because and only insofar as the SI prefixed units are ambiguous as to their meaning (decimal or binary?; c.f. User:Boatmurdered above and Kilobyte). As for losing the decimal approximations of the binary values, it makes the navbox rather wide and arguably would be better suited as actual article content (see Binary_prefix#Deviation_between_powers_of_1024_and_powers_of_1000). --Cybercobra (talk) 19:27, 6 February 2012 (UTC)[reply]

IEC, ISO and JEDEC Prefixes

[edit]

The IEC and ISO prefixes are documented in international standards for magnitudes up to and including YB and YiB. They are also used in hundreds of scientific articles every year. The JEDEC prefixes are documented in a JEDEC standard up to and including GB. Like compvis I do not believe the JEDEC prefixes are used for large magnitudes. The version proposed by compvis is better than that proposed by arthur rubin. Dondervogel 2 (talk) 12:27, 14 April 2013 (UTC)[reply]

Merriam-Webster defines terabyte as "1024 gigabytes or 1,099,511,627,776 bytes" first and "one trillion bytes" second. My personal belief is that being concerned with using every available byte is infeasible beyond the megabyte level, so the distinction between the powers-of-ten based meanings and powers-of-two based meanings becomes less important for quantities of one gigabyte and above. Hence usage examples tend to be more approximate and it will often be hard to discern which meaning the author of an example had in mind.
The dictionary I cited does not have definitions for petabyte or exabyte. Jc3s5h (talk) 13:51, 14 April 2013 (UTC)[reply]
The argument repeatedly used against IEC prefixes is that they are not used, so let's look at usage. WP's exabyte cites one IBM article that goes up to 16 EB with binary meaning. That's only one article but it seems an important one. There seems no justification to claim though, as this template does, that JEDEC prefixes are applied to quantities larger than 1 EB. Dondervogel 2 (talk) 14:09, 14 April 2013 (UTC)[reply]

See also: WP:COMPUNITS

[edit]

The template currently includes WP:COMPUNITS under "See also". I removed it on the grounds that "generally article space should not link to WP space", but Jc3s5h restored it, saying that "template space is not article space, and the link to the WP space page is not visible in the transclusion of this template".

While I agree that WP:Template namespace is not Wikipedia:Article namespace, and that the "see also" list is not visible (being tagged as <noinclude>), I still do not believe it is appropriate. Wikipedia:SELF#In the Template and Category namespaces says (with my emphasis)

Limited use of self-references are sometimes found in the Template namespace and the Category namespace, such as with disambiguation and stub notices. Expanding this to other areas is not encouraged...

If we really think that a reference to the WP page is necessary at all, it should probably be a {{selfref}} hatnote, but even then I don't think it appropriate or necessary. Mitch Ames (talk) 05:04, 23 November 2013 (UTC)[reply]

I consider the general practice in English in this area to be unstable. We have popular usage avoiding the IEC prefix, but standards-making bodies endorsing them and discouraging the binary meaning of SI prefixes. Thus, I anticipate that some sort of readjustment is likely in the future. Links between WP guidelines and templates will make it easier to be aware of the areas that should be examined if the practice in the English language changes.
Also, reading Wikipedia:Manual of Style/Self-references to avoid, I think it only applies to material that is visible in articles. If you look at {{cite book}}, for example, the documentation is chock full of links to pages in template and WP space. Jc3s5h (talk) 15:35, 23 November 2013 (UTC)[reply]

The International System of Quantities (ISQ)

[edit]

The so-called "IEC prefixes" are now part of the International System of Quantities (ISQ). Given that the ISQ is a broad collaboration between ISO and IEC, the heading "IEC" seems too narrow to describe the present situation. With this in mind I propose replacing the header "IEC" with "ISQ". Dondervogel 2 (talk) 11:29, 5 March 2014 (UTC)[reply]

Hm. As someone who generally prefers precise and current terminology, I concur. As someone who supports the notion of using the most familiar terms (e.g. WP:COMMONNAME), I have my doubts. Jeh (talk) 21:44, 5 March 2014 (UTC)[reply]
Fair comment. On the other hand, given that few readers have heard of IEC prefixes at all, even fewer will worry too much whether they are called "IEC prefixes" or "ISQ prefixes", so perhaps this is one of those times to favour precision. Dondervogel 2 (talk) 21:59, 5 March 2014 (UTC)[reply]

Styling using navbox classes

[edit]

A recent edit replaced in-line styling by navbox classes. Apart from an acceptable change of coloring, it also has the rather ugly effect that the main title is no longer centered, and the VTE temmplate no longer right aligned. I have tried to fix it, but could not succeed. User:Frietjes might have a try at correcting it. −Woodstone (talk) 16:55, 8 June 2015 (UTC)[reply]

should be better, we can't use width:2em due to the vte links floating outside of the sidebar on Firefox. Frietjes (talk) 17:11, 14 June 2015 (UTC)[reply]

It looks ok now. I suppose you will repeat the process on the sister templates of bits and prefixes? −Woodstone (talk) 15:39, 15 June 2015 (UTC)[reply]

Recent addition of terabyte in JEDEC column

[edit]

Are people comfortable with this edit? The issue is not whether the terabyte is sometimes used in this way (that it is has been established at Talk:terabyte), but whether said use merits it acquiring the same status as other entries in the same column. Dondervogel 2 (talk) 21:51, 4 August 2015 (UTC)[reply]

It's reliably sourced so no reason not to have it. It's irrelevant if someone personally dislikes it because it's what is reliably sourced that matters. The entire JEDEC column should should be renamed to "common prefixes" and contain all the commonly used prefixes as binary quantities because that's what is used in reliable sources. Again this is about correctly representing reliable sources and improving the coverage and relevant of Wikipedia. Fnagaton 23:44, 4 August 2015 (UTC)[reply]
Hello! Let's have a look at it from the current usability standpoint: where would TB be used as 10244 bytes, at least frequently enough? At the present state of technology, pretty much nowhere as the computer storage uses powers of ten (for both HDDs and SSDs), while single-computer RAM capacities in the range of terabytes are still very rare. Thus, IMHO we should be better without it. — Dsimic (talk | contribs) 06:49, 5 August 2015 (UTC)[reply]
Agreed. The "binary terabyte" can hardly be called a "customary usage" when there are no products that use it. And, as I wrote about three weeks ago in response to Fnagaton's arguments on another of these templates' talk pages: References for the entries in this column are here for kilo, here for mega, and here for giga. We can't say that JEDEC defines tera until this page exists, or until "tera" appears on this page.
These are the pages of JEDEC's dictionary in which they define the standard terms that they recommend to their member companies for use. Mere mentions of tera "in a binary sense" within JEDEC's other pages cannot be said to be at the same level, not until "the tera page" exists. Fnagaton has never responded to this point. I consider this point definitive, particularly as Fnagaton has never made a cogent rebuttal to it, only continuing to cite mere "mentions". Jeh (talk) 08:38, 5 August 2015 (UTC)[reply]
I agree with Jeh. Dondervogel 2 (talk) 10:57, 5 August 2015 (UTC)[reply]
Windows uses TB as binary for hard disks. The JEDEC also define it as a binary quantity. There is no reason not to include TB ain the JEDEC column. Glider87 (talk) 13:28, 5 August 2015 (UTC)[reply]
The "cannot be used at the same level" reason is WP:OR and does not prohibit other reliable sources being used. As I've done I've cited the other standards document that shows terabyte with binary quantities. Trying to claim a link must exist before you accept something is an invalid reason since it's not compatible with WP:RSGlider87 (talk) 13:41, 5 August 2015 (UTC)[reply]
The way Windows displays storage capacity is a rather good point, but how many times do we state the numbers Windows reports as the capacity of a storage device in a Wikipedia article, just to describe how large a device is? — Dsimic (talk | contribs) 13:49, 5 August 2015 (UTC)[reply]
This revert [1] does not follow WP:PAYWALL which says "Do not reject sources just because they are hard or costly to access. If you have trouble accessing a source, others may be able to do so on your behalf". The reference is accurate and the account is free to get. These other reliable sources also list all the prefixes with power of two values. [2][3][4][5][6][7][8][9][10][11][12] The last few also describe the power of two use as "common". Such common use means it gets included in the table with references and the column header can be changed to reflect this common use. I also don't have to respond to incorrect claims about fabricated links to pages that don't exist Jeh. You claim is not cogent Jeh because the JEDEC use of terabyte with power of two values is reliably sourced. Your insistence on a particular link being created is illogical Jeh and as Glider87 points out it is a violation of WP:OR and WP:RS. Fnagaton 14:44, 5 August 2015 (UTC)[reply]
Obviously a specific missing page is not by itself definitive, but the fact that "Tera" is missing from this page remains. No, you don't have to respond, in which case the point remains valid. I am not disputing that various JEDEC documents use the term. I am disputing that they define the term. Clearly, they do not. Jeh (talk) 20:14, 5 August 2015 (UTC)[reply]
(edit conflict) Fnagaton, IMHO pulling the WP:PAYWALL guideline as an argument for something that should be very simply accessible doesn't make much sense. — Dsimic (talk | contribs) 14:48, 5 August 2015 (UTC)[reply]
It is accessible and verifiable if you create a free account. There is a copy of it here without an account.Fnagaton 14:49, 5 August 2015 (UTC)[reply]
You can ask Swtpc6800 about that WP:PAYWALL guideline if you like.Fnagaton 14:53, 5 August 2015 (UTC)[reply]
On which page in that PDF is the definition of "binary" terabyte, please? Searching for "TB", "terabyte" and "1024" returned nothing. — Dsimic (talk | contribs) 14:56, 5 August 2015 (UTC)[reply]
Page 100 "The maximum density possible to be indicated is thus 2 Tera bytes (4 294 967 296 x 512B)" which is a power of value value. Fnagaton 14:59, 5 August 2015 (UTC)[reply]
Terabyte use as power of two values [13] and [14] with hard disks, from a hard disk manufacturer.Fnagaton 15:03, 5 August 2015 (UTC)[reply]
Wherein Seagate says " all major disk drive manufacturers use decimal values when discussing storage capacity." Doesn't exactly help your case. They do note that Windows uses "binary values" but no one has disputed that point. Jeh (talk) 20:14, 5 August 2015 (UTC)[reply]
As we know, this template tries to list various units as categories defined by the respective standardization bodies. Obviously, JEDEC doesn't define a "binary" terabyte; instead, JEDEC just mentions it somewhere and that's the key. Thus, if we include "binary" terabyte under the "JEDEC" column, it's no longer JEDEC who defines all units in that column. What should we list instead of JEDEC, as there must be some standardization body that defines various units? — Dsimic (talk | contribs) 15:51, 5 August 2015 (UTC)[reply]
I do not believe that any widely-recognized standardization body other than JEDEC has codified the use of any of the SI prefixes for any meaning other than as powers of 1000, and JEDEC has only defined kilo, mega, and giga. That's the whole point. The "JEDEC" column head is parallel to "Metric" (SI) and "IEC". Everything in the JEDEC column should be similarly defined by that organization. There is exactly one document at JEDEC where JEDEC publishes their standardized definitions of terms, and "tera" is not among their definitions. It is misleading to list "tera" in that column as if it were on an equal footing with the others. It is not "OR" to note that JEDEC's definitions document does not have a page for "tera", or that "tera" is missing from that document's list of terms that start with "T". On the contrary, it is blatant synthesis to generalize from a mere mention to a formal definition. Jeh (talk) 20:14, 5 August 2015 (UTC)[reply]
This template shows byte quantities, it's not only for what you thin is in standards bodies. It is misleading to not include significant points of view expressed in reliable sources.Glider87 (talk) 22:37, 5 August 2015 (UTC)[reply]
Jeh you are incorrect about the JEDEC only having binary tera in one document. The JEDEC JESD88E says "terabyte commonly used as a prefix to units of semiconductor storage capacity and meaning 240 [1 099 511 627 776] bytes" It is WP:OR to try to create any meaning about a link you create not existing, unless you can find a reliable source that supports what you claim about terabyte not being defined because of that specifc link you mention not existing. The mega entry in the JEDEC link already has an entry for tera with a binary quantity. So you are violating WP:NPOV by refusing to accept those reliable sources.Glider87 (talk) 22:41, 5 August 2015 (UTC)[reply]
Jeh your claim is incorrect because the JESD88E link above is from the JEDEC Dictionary of Terms for Solid-State Technology — 6th Edition. It is blatant synthesis to try to claim a dead link you CREATED means anything other than it's a dead link. Not updating one link in a dictionary in a particular part of a website doesn't invalidate common usage demonstrated by other links from the same organisation. As Glider87 correctly points out it's a violation of NPOV to not include commonly used prefixes in the binary column. Their common use is documented by reliable sources.Fnagaton 23:56, 5 August 2015 (UTC)[reply]
I don't understand the question about the JEDEC use of terabyte. This is not an extraordinary claim that requires multiple sources. It is just the next unit of size. I provided one clear expression of use in a JEDEC standard. (JEDEC Standard. MultiMediaCard (MMC) Electrical Standard, High Capacity (MMCA, 4.2) JESD84-B42 Page 100.) I was on the JEDEC Solid State Memories committee JC-42 in the 1980s. The standards are developed before the chips are produced, they may precede the actual devices by a year or more. The dictionary is maintained by JC-10. The June 2013 version of JESD88E is the current version and it defines a binary terabyte on page 226. The online dictionary has undue notability here on Wikipedia because of a foot note about possible confusion the decimal and binary use of megabytes on 3.5 inch floppy diskettes. -- SWTPC6800 (talk) 01:22, 6 August 2015 (UTC)[reply]
It is amazingly disingenuous of you - that is, of course, a polite way of saying "you're lying" - to cite that entry in JESD88E as support for the binary definition of terabyte. Did you think no one would check up on you? Here it is in its entirety, with boldface and font size changes preserved:

terabyte (TB) ( in reference to solid-state drive capacity): A memory capacity approximately equal to 1x1012 bytes.
NOTE Contrast with terabyte commonly used as a prefix to units of semiconductor storage capacity and meaning 240 [1 099 511 627 776] bytes.

The "commonly used" binary definition is in smaller type for a reason. JEDEC's primary definition, from the document Fnagaton or Glider linked, is the decimal one. The binary definition is clearly not intended to be taken as equally definitive. Even if we ignore the type size change, the best you could then say is that this definition gives equal weight to both definitions (but that would be OR, because the type size change is definitely there).
Meanwhile, the "tera" page is still conspicuously missing from their online dictionary. Fnagaton, you are correct, that a specific page link for tera, one that I made up, is not there is not definitive. But the fact that none of you will be able to find a page for "tera" alongside the ones for "kilo", "mega", and "giga" in the JEDEC online dictionary most certainly is: There is no page for "tera" in that dictionary. And well you all know it, no matter how you try to twist and turn to get away from it. Jeh (talk) 02:43, 6 August 2015 (UTC)[reply]
Fnagaton makes a good point that you've not refuted, namely that you cannot put any special meaning to an out of date dictionary when there are reliable JEDEC sources that do define terabyte with binary meaning. All you're doing is ignoring the evidence that shows you're wrong.Glider87 (talk) 15:10, 6 August 2015 (UTC)[reply]
My use of page 100 was the page in the PDF file, the printed page is 86. Here is an actual use of a binary terabyte.
JEDEC STANDARD "MultiMediaCard (MMC) Electrical Standard, High Capacity (MMCA, 4.2)" JESD84-B42 page 86. (PDF 100).
SEC_COUNT
The device density is calculated from the register by multiplying the value of the register (sector count) by 512B/sector. The maximum density possible to be indicated is thus 2 Tera bytes (4 294 967 296 x 512B). The least significant byte (LSB) of the sector count value is the byte [212].
It appears that JEDEC is using the binary Terabyte in their new standards. The working standards committees (such as JC-42) don't worry about what is or isn't in the JEDEC dictionary. I was on JC-42 for about 10 years. --SWTPC6800 (talk) 04:03, 6 August 2015 (UTC)[reply]
Frankly Jeh your attempts at fabricating links and what's actually used by the JEDEC is not neutral or reliable. You're wrong. So now you've been proven wrong with the facts, do you have any policy based objection to including all the commonly used prefixes in the table? Fnagaton 04:24, 6 August 2015 (UTC)[reply]
Guys, this is slowly becoming ridiculous. The fact that some JEDEC document mentions the "binary" terabyte doesn't imply that JEDEC defines it at the same time. For example, the JESD84-B42 document also mentions numerous aspects of the SPI standard, but doesn't define it. Also, the fact that Windows uses "binary" terabyte to report storage capacity doesn't mean that JEDEC had defined it. Thus, based on the references provided so far we simply can't conclude that it is JEDEC who defines the "binary" terabyte. — Dsimic (talk | contribs) 09:37, 6 August 2015 (UTC)[reply]
It's included in the standards document, which is a definition, and in the "Dictionary of Terms for Solid-State Technology" which defines terms used in the standards. As they say in the document "This Sixth Edition includes definitions from 14 standards that were not included in the Fifth Edition plus revised definitions from 26 additional publications and standards that have been updated. All reported errors and necessary rewording have been taken into account.". That is a JEDEC definition. Fnagaton 12:33, 6 August 2015 (UTC)[reply]
From which source is that quote, please? — Dsimic (talk | contribs) 12:37, 6 August 2015 (UTC)[reply]
It's in the foreword of JESD88E. Jeh is being silly trying to claim a difference in font size has some special meaning just for terabyte when it doesn't. In other sections we can see the notes have different font sizes. It's just layout style, nothing more. If he wants to prove it then he needs to produce a reliable source. Glider87 (talk) 15:04, 6 August 2015 (UTC)[reply]
I added reliable sources for the binary use of all these prefixes from the links provided above by Fnagaton at 14:44, 5 August 2015 (UTC). I could do with some help tidying up the references so they contain the book titles for each reference. Since the binary use of these prefixes is notable and common using reliable sources for each entry seems like overkill, but if you wanted to start referencing each use or definition then OK.Glider87 (talk) 15:34, 6 August 2015 (UTC)[reply]

Revert to 14 June 2015

[edit]

I reverted to the last stable version (14 June, by Frietjes) because

  • The table had become impossible to read because of all the references, and
  • the instability demonstrates a lack of consensus for the changes.

Editors are requested to make proposals here at Talk and gain consensus for changes before implementing them. Dondervogel 2 (talk) 22:30, 6 August 2015 (UTC)[reply]

See the section above for consensus. There are many reliable sources that show binary use so it should be included. You are requested to talk there. Glider87 (talk) 23:18, 6 August 2015 (UTC)[reply]
Removing the references and using this talk page as the repository of reliable sources would make the template more readable. The table has to include what significant reliable sources show. Removing that relevant information in that revert Dondervogel was the wrong thing to do.Fnagaton 00:02, 7 August 2015 (UTC)[reply]
As noted in my edit summary, there's absolutely no point in inundating numerous articles that transclude this template with all those references. Whatever we decide in the end, please let's keep the references on the talk page, not in the template. — Dsimic (talk | contribs) 11:15, 7 August 2015 (UTC)[reply]
@Dsimic Do you agreed that the recent additions are reliably sourced? If there is consensus for that, it is clearly time to make new articles for brontobyte and geopbyte as these are defined by the same "reliable source". Dondervogel 2 (talk) 11:29, 7 August 2015 (UTC)[reply]
I don't agree as those sources aren't definitions, but I'm trying to open a path for further constructive discussion that doesn't involve discussionless reverts. — Dsimic (talk | contribs) 11:38, 7 August 2015 (UTC)[reply]
All those links are definitions because they give the exact meaning of the words like terabyte etc. Reliable sources don't only have to come from standards organisations. Reliable sources have to be notable and they have to represent significant points of view. It is true that kilobyte all the way to zettabyte with binary values are notable and significant points of view. Glider87 (talk) 11:50, 7 August 2015 (UTC)[reply]
Are you aware that various units, such as kilogram, pound, meter or inch, must be defined by different standardization bodies? For example, see one of the updates to the definition of pound, issued by the National Institute of Standards and Technology (NIST). Thus, having some unit mentioned somewhere isn't automatically its definition. — Dsimic (talk | contribs) 12:06, 7 August 2015 (UTC)[reply]

@Dsimic The trouble with your approach is that the only thing Fnagaton and his sock do is repeat "you are wrong" over and over again, and that is not constructive. That is why I reverted to before the first disruptive edit - that way any proposed change is done from a stable basis, and not the present version, which I totally reject. Dondervogel 2 (talk) 12:38, 7 August 2015 (UTC)[reply]

You are wrong Dondervogel because you are not following policy and you are wrong because you resort to personal attacks about socks. For example according to Wikipedia policy and you can read the policy for the reasons why. Specifically articles and this template that is used in articles must be seen to be "fairly representing all majority and significant-minority viewpoints published by reliable sources" which means including "terabyte/megabyte/petabyte/etc = this power of two value". The definitions for this use comes from reliable sources and all the links provided are considered to be reliable sources for the purposes of defining the common usage of these prefixes. If you disagree then please cite precisely the Wikipedia policy that says only standards organisations definitions must be used in articles? If you cannot then provide that cite then there isn't any basis for excluding this common usage. It's not simply "being mentioned" it's multiple reliable sources reflecting real world usage and defining "terabyte/megabyte/petabyte/etc = this power of two value". As Swtpc6800 pointed out above "This is not an extraordinary claim". Not including this usage is not following the WP:NPOV policy. You hyave no policy based reasons for rejecting the good change to the template that ihcludes this extra reliably sourced information Dondervogel. Your edit can be considered disruptive because your revert pushes a point of view that is contrary to policy. Fnagaton 12:43, 7 August 2015 (UTC)[reply]
Yeah, so we have that "you are wrong" once again. :) Fnagaton, regarding the guideline you're asking for, it's covered by what WP:SKYISBLUE talks about, simply because the fact that various units and prefixes are defined by different standardization bodies prety much belongs to the elementary school knowledge. — Dsimic (talk | contribs) 12:51, 7 August 2015 (UTC)[reply]
So you agree that the standard for inclusion into articles is "fairly representing all majority and significant-minority viewpoints published by reliable sources" not "must be defined by standards organisations". This means there is no good reason not to include "terabyte/megabyte/petabyte/etc = this power of two value" in this table. This is because it's neutral to include this information. The point of view that is being fairly and properly rejected is that from Dondervogel because his point of view that this reliably sourced common use should not be included in the table, his point of view is not neutral. Fnagaton 13:17, 7 August 2015 (UTC)[reply]
From WP:SKYISBLUE "show that the presentation of material on Wikipedia is consistent with the views that are presented in scholarly discourse or the world at large". What shows the the views that are presented in the world at large? Is it to show the definition and use of these prefixes with binary values? Or is it hiding the facts? The answer is of course showing the definition and use of these prefixes with binary values which means the common use column needs to be filled with all the prefixes supported by reliable sources, not just kilo/mega/giga. Definitions come from other sources not just standards organisations. Wikipedia is full of definitions from these kinds of reliable sources. Common use defines these binary prefixes, they are ipso facto definitions. Trying to say that definitions only come from some standards organisations ignores the world at large.Glider87 (talk) 15:19, 7 August 2015 (UTC)[reply]
@Fnagaton: This isn't an article, this is a template and as such it doesn't establish a NPOV; instead, it just provides links to other articles, and a summary of information provided by other articles. Thus, this discussion shouldn't be taking place here at all, but that's what we have at our hands. The primary question, I'll repeat it again, is do we want to include only formally defined "binary" units, or all commonly (and less commonly) used ones? Though, you don't seem to be willing to truly discuss the whole thing in a constructive way.
@Glider87: Technically, it's wrong to use SI prefixes (kilo, mega, etc.) to represent any other meanings than their SI-defined meanings of powers of ten; "binary" uses of those prefixes are an exception that's defined by another standardization body, which is JEDEC. I don't see the point of discussing the whole thing with you, as you seem to refuse to take any other arguments into consideration, and the primary argument you should consider is that various units and prefixes must be defined by respective standardization bodies. — Dsimic (talk | contribs) 20:25, 7 August 2015 (UTC)[reply]
I've just had a lovely ploughman's lunch and a decent local brewed beer in a little west country pub, so I'm being generous when I say it really doesn't matter if you think it's "technically wrong" because the majority of reliable sources regarding this topic use these prefixes with binary uses. Your arguments, which are weak, do not allow any of us to ignore the policies and guidelines about reliable sources and neutrality. Neutrality means we all have to reflect use as the reliable sources show us. Glider87 (talk) 12:11, 8 August 2015 (UTC)[reply]
Why would your lunch, beer you had, and the assessment that you're in a "generous" mood matter for this discussion? — Dsimic (talk | contribs) 13:21, 8 August 2015 (UTC)[reply]
I was in a good mood. Now this template has neutral and reliably sourced information in this template it's much better than before.Glider87 (talk) 15:40, 8 August 2015 (UTC)[reply]
Well, a person's mood shouldn't affect their viewpoints, especially when it's about establishing a NPOV. — Dsimic (talk | contribs) 15:44, 8 August 2015 (UTC)[reply]
It's the facts that are relevant for establishing NPOV. So let's visit the facts. The prefixes all have many reliable sources for binary use. Such use has a long history and is common in reliable sources. What possible neutral explanation is there for not including such commonly used binary prefixes? Glider87 (talk) 16:03, 8 August 2015 (UTC)[reply]
I've already explained it a few times, and there's point in repeating it all once again. — Dsimic (talk | contribs) 12:37, 9 August 2015 (UTC)[reply]
  • This template is used to provide common information displayed inline in articles. It is therefore subject to the same neutrality policy as the articles. Since articles have to include commonly used binary units then this template has to as well. In answer to your question, it doesn't matter what you want to include because those guidelines and policies mentioned above obligate you and everyone else to be neutral and include these binary units because that's what the reliable sources do. See [WP:MOSNUM] for the reasons why.Fnagaton 21:06, 7 August 2015 (UTC)[reply]
    It's a shame that people don't seem to want to talk about Wikipedia editing policies and instead revert without comment. They're not really trying to build consensus it seems.Glider87 (talk) 15:44, 8 August 2015 (UTC)[reply]
    If they don't talk they're not engaging in the consensus process. Reverting workout talking would be disruptive.Fnagaton 16:59, 8 August 2015 (UTC)[reply]
  • I'm an uninvolved editor. It's clear to me that the long list of references should not be included because it makes the rows no longer line up. It's also obvious from watching the long lame back-and-forth that there is no consensus to add anything to the JEDEC/Common Use column beyond "giga". Reyk YO! 09:03, 9 August 2015 (UTC)[reply]

An extra column can show the relative difference between Decimal and Binary

[edit]

Please have a look at https://nl.wikipedia.org/wiki/Veelvouden_van_bytes It has a third column:

  • 2.4% difference between kB and KiB
  • 4.9% difference between MB and MiB
  • ...
  • 10.0% difference between TB and TiB

--> Consider adding such third column in the English version too. — Preceding unsigned comment added by 162.74.52.147 (talk) 14:21, 4 July 2019 (UTC)[reply]

Good suggestion. See also Binary_prefix#Deviation_between_powers_of_1024_and_powers_of_1000. Dondervogel 2 (talk) 14:35, 4 July 2019 (UTC)[reply]
Oppose I also agree with Woodstone that this is not the right place for a comparison. The comparison can probably be better explained in an article. Somerandomuser (talk) 04:56, 29 July 2021 (UTC)[reply]

Recent Edit to Table

[edit]

Please leave as is - there does not appear to be an reference to the JEDEC Binary TB. this column should be left blank to avoid additional confusion. As previously discussed, JEDEC JESD100B.01 [1] falls short of defining the TB, instead, referencing IEC 60027-2. An alternative system is found in Amendment 2 to IEC 60027-2: Letter symbols to be used in electrical technology – Part 2:

Further, JESD218 Clearly defines the TB as "a terabyte is equal to 1⋅1012 bytes." "For the purpose of this standard", which is clearly the decimal base 10 representation of the TB [2] (section 3.2.1.)

@Dondervogel 2 rolled the change out - I agree - leave this blank.


EDIT/UPDATE The Binary Terabyte represented by unit TB is found in the March 2016 edition of Revision of JESD220-1[3] on Page 2 where TB is defined as a Terabyte AND on page 3 where Terabyte: 1.099.511.627.776 or 240 bytes.

Open for discussion? User:Thegiz 28 April 2021

That's interesting. Are you saying that JEDEC defines the terabyte as both a decimal quantity and a binary one? Can you provide a little more context for both definitions? Is there any one JEDEC document that defines all of KB, MB, GB and TB? Dondervogel 2 (talk) 10:33, 29 April 2021 (UTC)[reply]
IIRC, the JEDEC docs do not "define" the units. They list existing definitions and issue recommendations for use in the industry branch and indeed they acknowledge, even recommend, the metric standard, but permit the binary versions as alternative use by vendors. IMHO, the JEDEC column should just be removed. kbrose (talk) 01:19, 5 May 2021 (UTC)[reply]
But JEDEC does define prefixes K, M and G for semiconductor storage capacity. It does not define T. With this in mind, I think the JEDEC column should stay, but limited (as now) to the 3 prefixes that are defined. Dondervogel 2 (talk)
The JEDEC documents explicitly state that they list the binary interpretations only because of common usage, not as legal requirement or binding definition. You can't take the output of this form query as anything but a citation. Reading the full documents reveals the true intent of the documents. They in fact acknowledge the decimal standards. If a manufacturer wants to label products with binary prefixes, they are fully within the specs. IIRC, this is even the recommended practice, but for historical reasons manufacturers are free to use binary interpretations of metric prefixes. If anything, the JEDEC column should list both. Better is not to list them at all, because they do not promulgate a standard. It is a listing like any other in many reference works. They only listed those prefixes that were actually used in products at one time, they may have indeed added T since, because now terabyte memories exist. This practice shows that these are not definitions, but listings of common usage. kbrose (talk) 12:16, 5 May 2021 (UTC)[reply]
There is no entry for T in JEDEC's dictionary of terms (JESD88). Dondervogel 2 (talk) 12:45, 5 May 2021 (UTC)[reply]
Maybe not. But this is meaningless, really. Listings are not prescriptive. There is no context to any of those. You have to read the text. Here is what is written in 100B-01, cited by the editor above:
The definitions of kilo, giga, and mega based on powers of two are included only to reflect common usage. IEEE/ASTM SI 10-1997 states “This practice frequently leads to confusion and is deprecated.” Further confusion results from the popular use of a “megabyte” consisting of 1 024 000 bytes to define the capacity of the familiar “1.44-MB” diskette. An alternative system is found in Amendment 2 to IEC 60027-2: Letter symbols to be used in electrical technology
So, JEDEC explicitly states that the binary interpretation of the prefixes is DEPRECATED. What else does one need to make the point? kbrose (talk) 13:05, 5 May 2021 (UTC). PS: The editor apparently has access to a newer draft of the document, I can only access the 2002 version right now.[reply]
Standard 218 (2010, SSD drives) states: For the purpose of this standard, a terabyte is equal to 1⋅10^12 bytes kbrose (talk) 13:19, 5 May 2021 (UTC)[reply]
Standard 220 (2013, for UFS) lists: Terabyte: 1.099.511.627.776 or 2^40 bytes.
So, it depends entirely on specific product classes which definitions are listed, because each has a common usage history. Therefore, one cannot just pick one usage and list it as the definition that JEDEC uses, as is purported in our table. The fact is that JEDEC does not prescribe definitions for units, but lists common usage only. kbrose (talk) 13:27, 5 May 2021 (UTC)[reply]
I guess we need some input from other contributors to this Talk page. Pinging @Woodstone: @Dsimic: @Jeh: @Swtpc6800: @Jc3s5h: @Cybercobra: @Quantling: @Thegiz:. Dondervogel 2 (talk) 13:25, 5 May 2021 (UTC)[reply]
The JEDEC column does not belong into this table. It is a column of exemplary usage, nothing else. By the same token, anyone could justify adding a column for Microsoft showing their use of the units in Windows, or someone could add a column for Seagate to show decimal usage, and cite some manual as definition for each. JEDEC does not set standards for units, but for products. The table is for the standards-based definition of the units, not examples. kbrose (talk) 14:52, 5 May 2021 (UTC)[reply]
I understand the point about giving undue weight to JEDEC, but I feel we should establish a new consensus before making a change like that. Before making a decision, we can also consider how well definitions from other standards bodies (ANSI, ISO) align with IEC. Dondervogel 2 (talk) 16:03, 5 May 2021 (UTC)[reply]
The only thing being given undue weight in this is the IEC units. Unlike those, the JEDEC units are actually used in marketing, documentation, reviews and packaging for memory-related products. —Locke Coletc 16:09, 5 May 2021 (UTC)[reply]
For reasons not clear to me this issue appears to be very sensitive. Changes will likely lead to edit wars. Therefore I am in favour of keeping the status quo.−Woodstone (talk) 16:22, 5 May 2021 (UTC)[reply]

Do we have a user insisting on using those ‘kibibit’ units again??? We should only use units of measure that are predominately used in the industry for communicating with a regular customer base. It’s not complex. To those who respond that “we need to disambiguate,” I agree. So, how is it you can’t figure out a way to disambiguate using conventional units of measure everyone is familiar with, just like the big computer manufacturers do? Greg L (talk) 23:27, 7 May 2021 (UTC)[reply]

I find the table misleading. It suggests that wherever JEDEC applies, the binary meaning is to be interpreted, or that JEDEC mandates this interpretation to the extent that it might be regarded as a standard. Unless we can negate this perception, I feel strongly that the JEDEC column should be removed. Clarifying this might be done by heading the column "JEDEC (deprecated)", or by replacing "JEDEC" with "Common usage". Otherwise, this column simply violates the encyclopaedic mandate. —Quondum 13:56, 14 May 2021 (UTC)[reply]
I second this. The column should be removed. JEDEC does not define (in the sense of setting a standard) the units at all, and does not mandate use of any specific units. Their documents explicitly state that units are listed to reflect common usage, but point out that the binary meaning is deprecated by standards bureaus and is ambiguous. kbrose (talk) 15:39, 14 May 2021 (UTC)[reply]
Yeah, no... we go by what our sources say, and our sources don't use KiB/MiB/GiB/TiB or their longform variations, as you can see at IEC units are bad. If anything, the column for the IEC units should be removed as it is giving undue weight to a "standard" that has nearly zero adoption by the industry, scholars, schools, retailers or marketing experts. We do not use terms just because someone makes them up and hopes we'll all adopt them. —Locke Coletc 17:04, 14 May 2021 (UTC)[reply]
@Quondum: The column is called JEDEC because that is the organization that has continued to use the classical kilobyte/megabyte/gigabyte/terabyte terms despite IEC units being a "standard". Because they use those terms the manufacturers of RAM/memory continue to perpetuate their use in brochures, marketing, packaging and in specifications. Other organizations also use these classical terms, such as Apple and Microsoft in both their technical literature as well as in marketing. Literally the outlier here is the IEC units, which, as detailed at IEC units are bad, are not widely used in our sources or by the public at-large. —Locke Coletc 17:04, 14 May 2021 (UTC)[reply]
What sources say that JEDEC standardizes KB, MB and GB in the binary sense? I am not objecting to having the column per se, only to the current heading as used with this column. What we can't have a template that implies what we can't source, namely that the JEDEC standardizes this meaning. —Quondum 17:23, 14 May 2021 (UTC)[reply]
JEDEC has an online dictionary that defines "kilo", "mega" and "giga" in the binary sense, in the context of semiconductor storage capacity (perhaps that restriction should be made explicit). It does not define "tera" or larger. Dondervogel 2 (talk) 18:07, 14 May 2021 (UTC)[reply]
It LISTS those prefixes because of history of common usage, as detailed in the listing of mega, to which giga refers. The reason that tera is not listed in 1999 is clearly that it had no common usage in memory products at the time. This makes it clear that these listing are not standards definitions for industry, but examples, just like Microsoft might list their use of units or any other company. In other product specifications in years later, tera was listed with both definitions, depending on product history. Their usage pattern makes it clear that they had no intention to set any standard, which is exactly why they refer to other standards (SI & IEC) in the giga entry. kbrose (talk) 19:28, 14 May 2021 (UTC)[reply]
Dondervogel 2, I had that "dictionary" in mind when I made that point. I agree with kbrose's counterpoint: that is very much JEDEC avoiding setting a standard (or even directly endorsing the use). However, I agree with you that it should be made clear the use of "common use binary prefixes" are used in a restricted to limited class of quantities within those with unit byte. Attaching JEDEC's name to this common use is just spurious. —Quondum 21:26, 14 May 2021 (UTC)[reply]
I guess if the "JEDEC" header were replaced with "conventional/non-standard use for semiconductor storage capacity", one could then justify including TB in that column, but that's one heck of a mouthful. It would need to be abbreviated somehow. Dondervogel 2 (talk) 09:03, 15 May 2021 (UTC)[reply]
Maybe just "Memory"? In a table, the brevity would be understood well enough. Personally, I'd prefer to remove the column, but there are some who seem to think that WP is supposed to reflect common usage (it isn't). It is interesting to see that a niche use (semiconductor memory sizes) qualifies as dominating all usage of units, and that all other products and uses do not count under common usage either. —Quondum 17:21, 15 May 2021 (UTC)[reply]

Suggested options

[edit]
I agree the "JEDEC" definitions do not have the status of ISO/IEC 80000, but I do not see a consensus here for removing the column. On the other hand I don't see a consensus for the status quo either. Options include
  1. No change
  2. Change "JEDEC" heading to "convention/memory" or similar, leaving KB, MB and GB only
  3. Change heading per #2 and add TB
  4. Remove JEDEC column
Have I missed any important ones? Dondervogel 2 (talk) 17:38, 15 May 2021 (UTC)[reply]
"Memory" is a better choice than "JEDEC", and I would add a foot note or second line (deprecated) to express the sentiment of JEDEC on this issue. JESD100 can even be used as a reference for that. kbrose (talk) 17:46, 15 May 2021 (UTC)[reply]
The list seems to sum it up. The only option that I feel is untenable is #1 (no change). —Quondum 01:12, 16 May 2021 (UTC)[reply]
I don't see how we can just omit the "JEDEC" column (option #4). After all, the convention described in that column is in widespread use. Further, if we do not associate the binary use with JEDEC, there is no longer a justification for stopping at GB (discounts #2). That would suggest we are left only with option #3, except that Woodstone clearly favoured the status quo (#1). One problem I see with both #2 and #3 is that the heading "Memory" seems too vague. We also need to convey that it is a widespread and non-standard convention (for computer memory). Can someone suggest a suitable header? Dondervogel 2 (talk) 09:56, 20 May 2021 (UTC)[reply]
I think Woodstone is simply observing that WP has editors who are somewhat dogmatic and combative about this, which I interpret as crossing only #4 off the list for the present to keep the peace. Since I am objecting to #1, and if the heading changes to reflect that this is related to usage, this means that #3 might be what could be targeted. I think of template as a reference: here, a kind of key of notable notations and their links. The first two columns are unassailable: they are supported by standards, they have widespread use in many areas other than in restricted areas. These "restricted areas" seem to me to be (i) binary-addressable memories that are expected to be mapped into a larger address space, and (ii) some perverted leakage of the SI-as-binary use into other areas, which includes a few legacy floppy disk formats and possibly a few other isolated cases (though SI prefixes dominate for memory sticks, disk drives, etc.), technically illiterate applications and OSes (Windows included) and a few insane court challenges. I'm guessing that JEDEC's usage falls into (i). Microsoft's use in Windows influences a lot of people, but this is no reason to adopt this use in WP. As an example, we don't drop spaces before units just because doing so is very common – we don't even give it a mention other than that it is not SI-compliant. If we find that 'TB' actually is used in the binary sense for (i) (it might be, in supercomputers), or if we assume that (b) is to include (ii) (its known use is likely for file/disk sizes only in the UI of Microsoft and a few others, becoming fewer), then adding TB makes sense anyway; I see no harm in doing so.
TD;LR suggestion: choose a new heading, and give it mouse-over explanation to deal with the constraints of brevity and include 'TB' in the column, i.e. #3. The heading could be 'Memories', 'Addressable', 'RAM', 'Historical' or similar, with a mouseover saying 'Often used for stackable addressable memory sizes.' The mouseover text could be elaborated somewhat if desired. —Quondum 15:32, 20 May 2021 (UTC)[reply]
I like the mouse-over idea. With that in mind, my preference is now #3. (And WP does sometimes drop spaces between value and unit symbol. An example is "5%") Dondervogel 2 (talk) 09:52, 21 May 2021 (UTC)[reply]
Well, yes – WP's style for quantities and units is based on the SI, but deviates from it, so we do not space '%'. WP is probably also nonstandard on currencies. Unfortunately, some areas of WP MoS guideline are looser, it seems largely because these areas provoke strong feelings in the editors. WP:ENGVAR is an example, where nationalism seems to have gained a foothold, creating a bizarre array of jealously protected dialects in en.WP, growing in number. Date formats are another. Unfortunately, this "historical data units" seems to be another such area. I'm glad that WP:LQ seems to have settled into a definite single style. I'll give it a first attempt, and see what the reactions are. —Quondum 12:38, 21 May 2021 (UTC)[reply]

Looks good. I added TB. Also, let's not forget

Should they not be made consistent? Dondervogel 2 (talk) 15:20, 21 May 2021 (UTC)[reply]

Each case is different
  • Template:Bitrates: IMO, absolutely not. I don't believe that the "traditional binary" prefixes have become the norm anywhere for communication rates, though it is possible that broadband advertising may be a bit strange. Where binary prefixes are convenient, the IEC prefixes must perforce be used to avoid confusion. The field of communication is not tolerant of imprecision: you need the figure to be precise and unambiguous.
They are not the norm for rates, but can nevertheless be found quite frequently in many networking tools and applications that had traditionally used the binary sense of the metric prefixes. kbrose (talk) 00:44, 22 May 2021 (UTC)[reply]
Quondum 18:07, 21 May 2021 (UTC)[reply]
  • Bitrates: I agree
  • Quantities of bits: SDRAM uses Mbit, Gbit, sometimes with a binary meaning and sometimes decimal.
  • Bits and bytes: Let's revisit once we've decided what to do with Mbit, Gbit
Dondervogel 2 (talk) 20:59, 21 May 2021 (UTC)[reply]
Okay, I'll take your word for some classes of memories being sized in bits. Which suggests that what has been said about bytes carries over to bits: I cannot see manufacturers being that different in the use of prefixes if they deal in both units. Us treating byte and bit the same here would simplify things. —Quondum 22:21, 21 May 2021 (UTC)[reply]

Plural column heading

[edit]

Using memories, the plural, in the column heading is rather odd syntax for a device class memory. Why? kbrose (talk) 20:12, 21 May 2021 (UTC)[reply]

The headings 'Metric' and 'IEC' refer to contexts, not to things that are measured. They are both evocative of their meanings (e.g. representing the phrases 'as a metric prefix'/'in the context of the metric system' and 'as standardized by the IEC'/'in the context of IEC standards'). In a similar vein, 'memories' is suggested by 'in the context of sizing memories', and not 'in the context of a memory device'. However, there are phrases that fit 'memory' too: 'in the context of memory sizing'. So it is really a coin flip. I had to choose something, and at the time, that seemed to fit. I don't really care either way. I was sort of hoping someone would come up with something quite different that would fit better. —Quondum 22:06, 21 May 2021 (UTC)[reply]
The phrase sizing memories is also unusual. It would be appropriate for amounts of memorable happenings (memories of a life time). Sizing memory is most likely much more common in information technology. A set of RAM modules is not called memories, just memory, and that even applies to a farm of disk drives. kbrose (talk) 00:33, 22 May 2021 (UTC)[reply]
Just saying how my head worked at the time. Feel free to change it as you wish (but not 'JEDEC'). I wasn't defending my choice, only trying to answer the question with an insight of how I happened to settle on it. —Quondum 00:53, 22 May 2021 (UTC)[reply]

Deprecated

[edit]

From the linked PDF:

NOTE 2 The definitions of kilo, giga, and mega based on powers of two are included only to reflect common usage. IEEE/ASTM SI 10-1997 states “This practice frequently leads to confusion and is deprecated.” Further confusion results from the popular use of a “megabyte” consisting of 1 024 000 bytes to define the capacity of the familiar “1.44-MB” diskette. An alternative system is found in Amendment 2 to IEC 60027-2: Letter symbols to be used in electrical technology – Part 2:

JEDEC is not deprecating the term, they are quoting IEEE/ASTM SI 10-1997 (hence those weird symbols around This practice frequently leads to confusion and is deprecated.; those are quotation marks). You're attributing the IEC deprecation of the terms to JEDEC, which JEDEC is just documenting the issue as a NOTE in their standard. —Locke Coletc 06:55, 21 June 2021 (UTC)[reply]

JEDEC is stating the official status of these definition, that they ARE deprecated by the standards organizations. JEDEC does not define or recommend units, and does not bypass or dismiss the standards orgs. Their language of citing common usage is clear. The same still common, but deprecated usage for memory is now properly reflected in the table. kbrose (talk) 13:57, 21 June 2021 (UTC)[reply]
@Kbrose: Uh... that quote has been available since 1997 since the "standard" was first produced. JEDEC notes the issue and quotes the reason, but is not deprecating the terms for their definition. In addition to the PDF above, this definition is also publicly available on the web at this URL. You can see they still define "mega" as the traditional meaning even with their note about the IEC unit.
I also find it amusing that you've even quoted this passage before six years ago and are just now acting like it says that JEDEC "deprecated" the term. —Locke Coletc 17:19, 21 June 2021 (UTC)[reply]
Binary use of those symbols is deprecated by multiple standards bodies, including BIPM, IEEE, ISO, and NIST. Dondervogel 2 (talk) 17:24, 21 June 2021 (UTC)[reply]
The quoted passage is from 1997. How deprecated is it really if everyone is still using it...? And JEDEC is the organization that releases memory standards and has defined these terms in this way, terms which manufacturers, system builders, software developers and so on still continue to use. —Locke Coletc 17:43, 21 June 2021 (UTC)[reply]
Does the date really matter? If you insist on a date audit, the deprecations I am referring to are from 2008 (NIST, SP811), 2009 (ISO 80000-1), 2016 (IEEE, SI 10) and 2019 (BIPM, SI Brochure). Dondervogel 2 (talk) 18:04, 21 June 2021 (UTC)[reply]
The quoted passage being used to justify "deprecated" is from 1997. Can someone explain what changed in the past month to warrant noting this in the template when the term is clearly still being widely used by the industry at-large? —Locke Coletc 18:07, 21 June 2021 (UTC)[reply]
I believe the edit that sparked this thread was this one from 28 Apr. The subsequent discussion led to the present consensus. Dondervogel 2 (talk) 18:25, 21 June 2021 (UTC)[reply]
OK, so nothing changed. We're safe to revert back since apparently you guys misunderstood the JEDEC source discussed above that was used to support this change. —Locke Coletc 18:58, 21 June 2021 (UTC)[reply]
The JEDEC document is clear and does not support your advocacy and bias. JEDEC does not "define" units, as you wish to assert, they clearly state that they only list common usage, no matter that it is deprecated, which they note. It is similar to Wikipedia still permitting the binary interpretation because of common usage. Nobody here disputes that this usage still exists, but countless software has been converting and the new units are commonplace in new software. BTW, I have objected to the column "JEDEC" in these table since introduction, since it is misleading, but contrary to some members of the opposition, have not engaged in endless edit warring to assert my view, just like many others who are of the same persuasion that WP, too, should follow standards organizations. Every other conceivable technical detail is routinely updated when the standard orgs change definitions, not so for these units. It is time that WP follows the trend. kbrose (talk) 20:31, 21 June 2021 (UTC)[reply]
The JEDEC document is clear and does not support your advocacy and bias. Bias? Advocacy? I'm just going by what our sources say, and in this instance, what JEDEC says. JEDEC does not "define" units And yet they literally have an online "dictionary" (which, if you go back to one of the letters, clearly marks itself as a "dictionary", and even has "dictionary" in the URI). It is time that WP follows the trend. Perfect, can you please demonstrate this "trend"? WP:ECREE. Exceptional claims require exceptional sources per WP:V which is policy. —Locke Coletc 20:56, 21 June 2021 (UTC)[reply]
  • Follow the common practices of the RSs and the computer manufacturers It doesn’t much matter if any particular standards organization recommends this practice or that. What matters for a general-interest encyclopedia is to follow the practices most commonly observed and widely recognized in the industry so as to not confuse our readership. Wikipedia ought not be exploited in the hopes of leading the computer world to adopt new terms and units of measure. Our readership doesn’t know what JEDEC and IEEE/ASTM SI 10-1997 are; when we quote what one of these bodies is recommending, we're looking in the entirely wrong places for guidance. Likely over 99% of our readership has zero idea what a “gibibit" is because it’s obscure terminology; maybe that will change one day. And when the computer world changes, then Wikipedia can change, but not before. Greg L (talk) 01:45, 23 June 2021 (UTC)[reply]
Please end this nonsense. People come to Wikipedia to learn topics they don't know all the time. By that argument WP should not have articles about topics they don't know, or use terms they don't know. That is what linking is for to lead readers to articles that explain the terms. Wikipedia is not leading anyone in using binary prefixes or metric prefixes in proper (standard) interpretation. The heading was labelled JEDEC, and per your argument that should not be used, indeed. It was changed, but for the reason that it is misleading, not that people don't know what it means. The computer world has changed well in advance of your acknowledgement, these units are now used in thousands of cutting edge applications and tools. WP has new editors coming here that want to use the units, because that is what is being used elsewhere today, but some refuseniks revert this and create more confusion than ever existed. As a result, many readers come here and find old nomenclature used, and repeat this in the media. It is a self-fulfilling prophecy. Tell people the lie often enough and they believe it, then justify the wrong by telling us that people don't use or understand the units. kbrose (talk) 18:52, 23 June 2021 (UTC)[reply]

Consensus

[edit]

Also, on the point of "there was consensus for this change" (paraphrasing), can someone who has revert warred me show that? Because I see in the discussion above that myself and Greg L were generally against changes that promoted the IEC units, and Woodstone was for maintaining the status quo because they predicted an "edit war"; and yet I see thee editors claiming "consensus". 3 to 3 is not consensus, and the fact that the rest of us left does not give you consensus (see WP:NOTSILENCE). And there is definitely not consensus to change other templates where this discussion was not held. —Locke Coletc 00:46, 22 June 2021 (UTC)[reply]

As there was not consensus for the changes made, we should restore the status quo revision until consensus exists for the memory units to be marked as "deprecated". —Locke Coletc 16:54, 23 June 2021 (UTC)[reply]

Which revision do you want to restore? Please request a specific revision by its oldid or UTC date/time. – Jonesey95 (talk) 19:17, 23 June 2021 (UTC)[reply]
There is no need to restore anything. The current version is the best compromise that ever existed for this table. It acknowledges that the binary use of units is still common and therefore still has that column, and it makes it clear the practice is deprecated by industry bodies (according to their own literature). References for that would be nice for completeness, but are the responsibility of the enclosing article.kbrose (talk) 00:19, 24 June 2021 (UTC)[reply]
Until consensus develops for this language it should be removed, as we would with any proposed change on Wikipedia. —Locke Coletc 16:38, 24 June 2021 (UTC)[reply]
Jonesey95, be aware that this request amounts to an attempt at circumventing the reason for the template editing restriction by the edit warrior who failed to get during the edit warring that triggered the restriction. In any event, it would just degenerate into a pointless argument about what would really be the "status quo" version. —Quondum 02:53, 24 June 2021 (UTC)[reply]
So... Greg L was supposed to "edit war" with you to prove the point? My edit warring was not helpful, but at least I was operating with the status quo in mind, you guys were acting like you'd established consensus for this when you hadn't. WP:NOTSILENCE. —Locke Coletc 16:38, 24 June 2021 (UTC)[reply]
Oh, yeah, and Greg_L just showed up here out of the blue, without any canvassing? Dondervogel 2 (talk) 17:12, 24 June 2021 (UTC)[reply]
@Dondervogel 2: Do you regularly make false statements, or are you just genuinely incapable of telling the truth? Honestly, very curious. —Locke Coletc 20:09, 24 June 2021 (UTC)[reply]
Jonesey95, 1029732001 is the oldid. —Locke Coletc 16:35, 24 June 2021 (UTC)[reply]
 Not done as far as the edit request, now that the page is protected - discuss any proposed changes any only activate an edit request if there is a consensus for a change. The page is most certainly at The Wrong Version, but that is not a reason for another revert. — xaosflux Talk 18:09, 24 June 2021 (UTC)[reply]
We should not be rewarding bad behavior (editing without consensus, then edit warring to maintain that edit). —Locke Coletc 20:09, 24 June 2021 (UTC)[reply]

Removal of column

[edit]

While I can understand that "deprecated" is not the right word to describe this column, I do believe that the removal of this JEDEC/memory column will partially resolve this issue. Not only does the column invite edit wars, it also creates confusion and ambiguity, as the prefixes conflict. If needed, I believe the JEDEC prefixes and this topic should be explained in accompanying article text instead of this table. Somerandomuser (talk) 04:47, 29 July 2021 (UTC)[reply]

For this proposed solution, we can keep the decimal and binary columns. The accompanying article text can be used to explain that JEDEC standards are binary despite using the metric/SI prefixes. Somerandomuser (talk) 05:12, 29 July 2021 (UTC)[reply]
Strong oppose. Unsupported by the sources, manufacturers, or major software vendors who still use the terms interchangeably, and definitely do not use the -ibi-bit/byte terms that IEC has been pushing since 1997.. the presence of -ibi units is actually the more misleading to our readers, as it suggests those units have seen widespread adoption (when nearly 0% of media sources use them, and less than 2% of sources overall use them). —Locke Coletc 05:24, 29 July 2021 (UTC)[reply]
Oppose: the template gives both the official and de facto prefixes to help the reader navigate the confusing situation. Both are well placed here. The only point of contention is the header of the binary K/M/G(/T) column. Proposals have been JEDEC, Memory, Traditional, Deprecated. Anyone has a better idea?−Woodstone (talk) 14:49, 29 July 2021 (UTC)[reply]
Customary, common, common usage, also used, also seen, also, Memory (customary), Memory (common usage), annoyingly normal, also OK, sick .... Actually, Memory (common usage) might not be too bad. NebY (talk) 17:13, 29 July 2021 (UTC)[reply]
Oppose: readers likely require some guidance to make out what's what. --Zac67 (talk) 17:00, 29 July 2021 (UTC)[reply]

Harmonise Bytes, Bits and 'Bits and bytes' templates?

[edit]

The bit and byte templates remain inconsistent, with 'Traditional', 'Memory' and 'JEDEC' labels used for what seems to me to be the same convention. I think the templates would benefit from further harmonisation. Dondervogel 2 (talk) 11:24, 15 August 2021 (UTC)[reply]

As has been discussed at Template talk:Quantities of bits, "Memory" suggests the terms were not used elsewhere when they were, and "JEDEC" is apparently objectionable for similar reasons (and also because some people think IEC is a more relevant standards body even though almost nobody follows them for their various -ibi terms). I'm not a fan of "Historical" because in some/most cases these terms are still used. —Locke Coletc 21:10, 16 August 2021 (UTC)[reply]
"Memory" seems to be the lesser evil or subjected to the least objection, so I went ahead and harmonized the templates (at least for now). --Zac67 (talk) 05:50, 17 August 2021 (UTC)[reply]

IEC units inclusion

[edit]

I note that IEC units are still being included in these tables despite no widespread use or adoption, this is, again, giving them undue weight and misleading our readers into thinking these terms are in common use. I think these tables could be much simpler with their removal, and instead a helpful header that notes where certain units are used (memory, storage, etc). —Locke Coletc 21:10, 16 August 2021 (UTC)[reply]

I've argued elsewhere that WP should not use IEC prefixes. But they are used occasionally in the real world. Therefore we should document them here, even though WP (and the real world) use them only rarely.  Stepho  talk  22:04, 16 August 2021 (UTC)[reply]
Oppose removing IEC prefixes. Even though only occasionally, they are in use in the real world as well as in WP. --Zac67 (talk) 05:50, 17 August 2021 (UTC)[reply]
they are in use in the real world as well as in WP By less than 2% of sources, a number I can count on one hand for press-related sources, and as far as I know few/any major software development shops use -ibi bit/byte in interfaces that are public-facing. Remember, the onus on inclusion rests on those wanting to include something; you don't get to vote your way out of WP:V. —Locke Coletc 07:08, 17 August 2021 (UTC)[reply]
Huh? Inclusion? I thought the discussion was about deleting. I look at the history and the IEC units have been there since the template was created in 2009. Which means it needs a consensus to delete them.
Also, there is some grey area between WP must use IEC units and WP must not mention them ever. No need to be black and white on this.  Stepho  talk  10:32, 17 August 2021 (UTC)[reply]
We don't generally include unsourced or poorly sourced information in Wikipedia, the fact that it has survived this long is not germane if the information violates other policies or guidelines. As to the grey area, for me that is having article-prose that mentions some standards bodies attempted to address the discrepancy but failed to gain any traction in the computing field at large. Prominently featuring these units, which are largely unused, in a table that features largely on pages including it is a tacit endorsement of a reality we don't live in. —Locke Coletc 07:55, 18 August 2021 (UTC)[reply]
Sheesh - you must be dredging the bottom of the barrel if you think we can't find references for the IEC prefixes. Here's 2 found in less than a minute:
https://www.iec.ch/prefixes-binary-multiples
https://physics.nist.gov/cuu/Units/binary.html
We don't normally put references in templates though. But the template does link to Binary prefix, which has plenty of references.
As for being rarely used - yes, I agree with that. But it's not a popularity contest. We document what prefixes can be used. Since they are rare, many people will not know what they are and will therefore turn to WP for help - making it even more important to document them. The same reason that we document many obsolete computers and computers that didn't sell very well on the market. We don't erase history and we don't erase unpopular facts.  Stepho  talk  10:39, 18 August 2021 (UTC)[reply]
@Stepho-wrs: I agree with most of what you say. I'd only take issue with the claim that IEC prefixes are rarely used. It just depends on where you look. For example they have been used in about 460 in scientific publications in 2020 alone. Is that so rare? When scientific authors choose to disambiguate, they do so almost universally using IEC prefixes [15] [16]. Dondervogel 2 (talk) 11:48, 18 August 2021 (UTC)[reply]
460 instances in 2020 is not a lot out of about 730,000 articles found using the same exclusions. It's not easy to compare because "GB" and "MB" turn up in so many irrelevant ways, but we can adapt your search (Google Scholar, 2020, various exclusions) to find instances of "256 MB", "512 MB", "256 GB" and "512 GB" - a very arbitrary selection of common IT values. Then we can run the same for MiB/GiB.
256/512 MB/GB: 5,840 hits.
256/512 MiB/GiB: 37 hits.
You might like to try that using other values such as 1, 2, 4, 64 and even 100 or 3, but I think you'll see similar ratios. NebY (talk) 12:28, 18 August 2021 (UTC)[reply]
@NebY: MB is far more common than MiB. No one here is disputing that. My main point is that MiB is used almost exclusively to disambiguate in scientific literature. Other forms of disambiguation are very rare (almost non-existent), by comparison. Dondervogel 2 (talk) 12:35, 18 August 2021 (UTC)[reply]
Or we're seeing that very many writers don't think they need to disambiguate or are doing it by other means. NebY (talk) 12:41, 18 August 2021 (UTC)[reply]
I provided 27 examples (of use of IEC prefixes for disambiguation) on memory hierarchy and 52 on SDRAM. I could have continued but I got bored - it was too easy to find examples. By contrast, only one counter-example was identified, of disambiguation on memory hierarchy. Dondervogel 2 (talk) 12:56, 18 August 2021 (UTC)[reply]
Google makes it easy to find examples of simple things like occurrences of GiB. Putting counts in context takes a little more work, as we see above, but if you don't do that a robust case can be made that results are being cherry-picked deliberately or by faulty methodology, as in the discussions you link. Yes, you didn't find a simple way to find if authors used different means of clarification and yes, you didn't examine how many authors don't find any need for such clarification or even deliberately avoid obscuring their meaning with excessive precision. Similarly, when you write "I'd only take issue with the claim that IEC prefixes are rarely used.... they have been used in about 460 in scientific publications in 2020 alone" and make "Is that so rare?" a rhetorical question rather than testing it yourself, you shouldn't be surprised if someone else finds the answer. Yes, it is so rare. NebY (talk) 16:10, 18 August 2021 (UTC)[reply]
Whether one considers 460 publications in one year to be few or many is a matter of opinion. As such, you are entitled to yours and I am entitled to mine, but to dwell on this is to miss the point about the near-complete absence of disambiguation methods other than MiB. I agree there are cases when it doesn't matter much, but there are also cases when it does. Can you provide examples of disambiguation of 2^20 or 2^30 bytes, in any publication outside Wikipedia (or WP clone), using something other than MiB or GiB? So far, editors have found only one counter-example (compared with 80 examples using IEC prefixes). i don't doubt there are more examples than just the one. I'm just saying they're hard to find. (You have offered none so far) Dondervogel 2 (talk) 18:30, 18 August 2021 (UTC)[reply]
We won't be using opinions on this, we'll be using facts. The facts do not support you. —Locke Coletc 06:58, 19 August 2021 (UTC)[reply]
No thanks - life's too short. The point is, again, that your evidence isn't sufficient for your claims that the usage isn't rare or that "when scientific authors choose to disambiguate, they do so almost universally using IEC prefixes". Think of it as basic hypothesis testing. NebY (talk) 18:19, 19 August 2021 (UTC)[reply]
Sorry if I wasn't clearer on this: I'm well aware that some standards bodies have created these terms. But we still do not give them undue weight when so few sources actually use them. See also: WP:NEO, another policy that pushing these terms violates (specifically the opening sentence of that section: Articles on neologisms that have little or no usage in reliable sources are commonly deleted, as these articles are often created in an attempt to use Wikipedia to increase usage of the term.). I don't think deletion is necessary, but as there are shades of gray to this, I also don't think unnecessarily pushing them is helpful either, our usage should reflect what our sources use. —Locke Coletc 06:58, 19 August 2021 (UTC)[reply]
@Locke Cole: There's no undue weight given by mentioning those prefixes – there's no pushing (at least not from my side). Also, their use for deliberate disambiguation seems to be fairly common, as can be seen above. Neologisms are something else, I don't think that applies here. --Zac67 (talk) 10:31, 19 August 2021 (UTC)[reply]
@Zac67: There's no undue weight given by mentioning those prefixes – I agree, so they can be omitted from this table, but mentioned in article prose on this subject as perhaps a "failure to adopt standardization" historical accounting. At this point, to me, it's more interesting (and more notable) how few companies/organizations use these terms than how many do. —Locke Coletc 18:01, 20 August 2021 (UTC)[reply]

Protected edit request

[edit]

Can the following please be added to the template as it has been nominated for deletion:

{{subst:Template for discussion|type=sidebar|heading=Quantities of computing information}}

Thank you. —Locke Coletc 17:32, 26 June 2021 (UTC)[reply]

 Done that, Locke Cole. Justlettersandnumbers (talk) 20:16, 26 June 2021 (UTC)[reply]
@Justlettersandnumbers: Please revert that edit now that the discussion has been closed as keep. * Pppery * it has begun... 00:58, 1 July 2021 (UTC)[reply]
 Done that, too, Pppery. Justlettersandnumbers (talk) 10:09, 1 July 2021 (UTC)[reply]

Bits and bytes templates

[edit]

We have two nearly identical templates but there are some differences between them. I have the following questions that seem to merit discussion

  • The Kbit/KB column is labelled 'Memory' in one and 'Traditional' in the other. Which is preferred?
  • Should there even be a 'Kbit' column at all?
  • Until a recent revert, the 'KB' column was deprecated in the bytes template. Is the KB notation deprecated or not?

Dondervogel 2 (talk) 12:13, 7 July 2021 (UTC)[reply]

  • The Kbit/KB column is labelled 'Memory' in one and 'Traditional' in the other. Which is preferred? See Template talk:Quantities of bits#JEDEC_column. Though upon reading the JEDEC definitions I'm starting to lean back towards JEDEC.
  • Should there even be a 'Kbit' column at all? Yes.
  • Until a recent revert, the 'KB' column was deprecated in the bytes template. Is the KB notation deprecated or not? No, it is not.
Locke Coletc 16:03, 7 July 2021 (UTC)[reply]
Several widely used operating systems report file sizes with binary units KB and MB. Therefore the header "memory" is too restrictive and "traditional" fits better. The usage of Kbit and Mbit is quite rare and does not merit to be in the template. Almost all official standards organisations declare the binary usage of K, M, G (and T) to be deprecated. That should be mentioned in the templates.−Woodstone (talk) 16:10, 7 July 2021 (UTC)[reply]
Therefore the header "memory" is too restrictive and "traditional" fits better. Pretty sure Dondervogel was referring to the bits column, not the bytes column. The usage of Kbit and Mbit is quite rare and does not merit to be in the template. Do you have any evidence of this? Almost all official standards organisations declare the binary usage of K, M, G (and T) to be deprecated. Given that most major manufacturers, software vendors, and even media outlets continue to ignore the deprecation by this standards body, why would we add Wikipedia's voice to something that clearly nobody is actually listening to? —Locke Coletc 16:24, 7 July 2021 (UTC)[reply]
@Dondervogel 2: I'll note that you seem to be ignoring the discussions above, as well as the discussion at Template talk:Quantities of bits#JEDEC_column. Is it your intention to start conversations over again anytime you don't like the way they go? Also, how do you reconcile your "questions" with the guidance provided at WP:COMPUNITS which says IEC units are generally not to be used? As this template is in article-space, it seems like WP:COMPUNITS answers your questions. —Locke Coletc 16:48, 7 July 2021 (UTC)[reply]
I'll take this as a "Yes". —Locke Coletc 07:22, 17 August 2021 (UTC)[reply]

New proposal

[edit]

A new proposal has been made at Quantities of bits. Please comment there. Dondervogel 2 (talk) 19:33, 10 November 2021 (UTC)[reply]

New prefixes

[edit]

Editors might like to join the discussion at Wikipedia talk:WikiProject Physics#Add ronna- and quetta- to units articles?. NebY (talk) 23:20, 18 November 2022 (UTC)[reply]