Hacker News Reader

Comment by XorNot | Hacker News Reader

Trying to figure out what pipe to ask for when I have the physical measurements in hand is one of the ongoing mysteries of the world for me. "Oh that pipe is <some number you will not get with any type of measuring tool>" is one of the most bizarre conventions out there.

That said, this isn't even the only place it turns up - there are timber sizes which are specified as one dimension and just "commonly known" to actually be a different one: https://www.thesprucecrafts.com/why-isnt-a-2x4-a-2x4-3970461

View on HN

Pipe was originally measured by inside diameter.

Anyway, over the past 100 years, metallurgy got better and they could make the same strength pipe with thinner walls.

So now, because of the thickness change, the old pipes that measured 1/2” inside and the fittings for those pipes were the right size for new pipes that measured about 3/5” inside. They had the same outside diameter, but a larger inside diameter.

Instead of confusing everyone and requiring them to remember compatibility charts for every size pipe (and guessing about the age of the pipe), they just “solved” the problem by selling the new pipes under the name of the old size they were compatible with.

Tl;dr: Plumbers have technical debt that goes way back.

Having seen the lead pipes Romans used, yes, plumbers have had some rough times.

The United States was still installing lead pipes until 1930-1986:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2509614/#__sec4...

They are called plumbers for a reason.

I know of at least one place which still had the original lead Roman piping quite late in the 20th century.

The timber sizes are different because you are buying planed timer, rough timber would have the exact dimensions you are looking for. So a 2x4 rough is really 2x4"

I’m in the throes of renovating an old house at the moment and dealing with replacing rotting sections of floor has proved a little fiddly due to this. The slightly different joist heights aren’t too bad - you just pack them out before strapping to the existing joists. Worse has been the existing chipboard floor, which is 20mm and I can only buy 18mm or 22mm. You can hack your way around it, but that one subtle difference adds quite a bit of time to the job - something I don think have much spare of at the moment!

You actually can get 20mm plywood, it will be a bit more expensive and likely much harder to find. Try a cabinet maker and ask if they can order it for you, cabinet makers have access to a vastly superior source of materials compared to your normal building store. Another option is that they route some grooves in thicker plywood at the spot of the joists (I guess that would qualify as a hack).

Thanks for the tip.

It’s irrelevant now because the flooring people have just walked out saying they won’t lay on to the chipboard anyway. Need to run a thin layer of ply over the whole thing. Live and learn I guess!

The timber sizes are different because you're buying planed dried timber. For some reason it was decided that measuring the timber wet and unplaned was a good idea.

On the other hand over here in metric-land if you buy a 44-by-88 piece of wood it'll be real close to 44x88mm, Whether you're buying it planed or not.

That's because historically it was cut then seasoned. What you ended up with a year later was somewhat smaller.

Everybody thinks that, but it isn't true. historically some sawmills where producing 2x4s that were 2x4, while others were using different sizes (I can't find evidence but I'm guessing they were trying to cheat their customers for profit and get away with it).

What matters isn't the actual size though, what matters is that when you are short a piece of wood you can get one that fits.

That's true, but they had to set their saws to something and a typical sawmill will process the logs as they're brought in, not after drying them because it would take forever to dry the whole log and it would likely split or be infested with creatures. So the logs are sawn 'green' to a rough size, dried (sometimes kiln dried to speed up the drying process) and then planed. Both drying and planing affect the dimensions.

It's quite amazing how the speed of the drying process can affect the later tendency of the wood to warp, as well as the degree to which it will warp. Counteracting this is quite tricky and requires all kinds of techniques, the most effective of which is laminating odd/even stacks of wood.

The same limitations hold in metric-land, and they still manage to label wood the size it actually is when you buy it. I'm sure it's a difficult problem but as far as I can tell it's also a solved problem. I imagine that the odd piece of lumber they have that dries with a severe warp or size deformity simply gets resawn.

Well, in 'metric-land' they don't have a history of naming their wood for the thickness of some guys thumb in integer multiples.

The USA could switch from 2x6 to 1.5x5.5 (planing takes 1/4" off each surface) tomorrow if they wanted to but I guess there would be quite a bit of confusion. The building industry is super traditional.

And when you do that you might as well go metric (for reference: 38x140mm).

> in 'metric-land' they don't have a history of naming their wood for the thickness of some guys thumb in integer multiples.

Yes we do. The metric system didn't fall out of a clear sky one day, it had precursors.

That's hardly the point though, even when using metric you could make the decision to measure wood wet. Even when using the imperial system you can make the decision to measure wood dry.

The sizes for timber come the actual physical size early in the production process as your link nicely explains. Where to the "commonly known as" sizes for pipes come from?

Technically HDD manufacturers are doing exactly the same.

Can you explain what you mean?

HDD manufactures often report a storage capacity that is larger than the effective capacity, since some capacity is used for formatting information, or error correction. This is the same situation with protocols (like gigabit Ethernet).

There's also "punning" between the SI préfixes, where e.g. "kilo" can be used to mean 1000 or 1024. See https://en.m.wikipedia.org/wiki/Kibibyte

Your second line is right. Your first isn't really. Formatting only reserves about 100MB. Error correction bits are completely hidden and not marketed at all. It's the 10% difference between "TB" and "TB" that really matters.

I think modern HDD and SSD manufacturers have been pretty honest about the actual capacity of their drives. They have no way of knowing what kind of filesystem you will be putting on the drive and how much overhead it has, or whether you might be putting it in a RAID 1 array (which cuts the effective capacity per drive according to the number of drives in the array).

The number of bytes they specify is the actual number of bytes available to your operating system. It seems like a pretty fair measurement.

The 1000 vs. 1024 thing is an unfortunate point of confusion. Terms like kibibyte and such were an attempt to solve it, but they never took off. (If you were to ask me which number a kilobyte or kibibyte was, I would have to go look it up!)

Even if we did use the kilo vs. kibi terms, HDD and SSD manufacturers are the ones who are getting it right. As the article you linked notes:

> 1 kibibyte (KiB) = 2^10 bytes = 1024 bytes

> The kibibyte is closely related to the kilobyte. The latter term is often used in some contexts as a synonym for kibibyte, but formally refers to 10^3 bytes = 1000 bytes, as the prefix kilo is defined in the International System of Units.

The same applies as you go up in the units. One mega-anything is 1,000,000 of those things, giga- is 1,000,000,000, and tera- is 1,000,000,000,000 things.

So a one-terabyte HDD or SSD should have a true capacity of 1,000,000,000,000 bytes, before any operating system or RAID overhead. Of course its actual physical capacity has to be higher, to support remapping of failing sectors or flash blocks and such. But that's all hidden by the drive controller.

I think it's the memory people who got this wrong, by co-opting a "kilobyte" to mean 1024 bytes, contrary to the standard definition. It was a handy coincidence of terminology at the time, but the error was amplified as we got into multiples of that size.

And then the operating system and utility people (or many of them) completely messed up by using the power-of-two definitions for disk/flash storage instead of the correct power-of-ten definitions.

This is why, for example, every single Amazon listing of a 1TB drive (HDD, SSD, flash card) which honestly provides the correct 1,000,000,000,000 bytes of storage to the OS will have at least one review complaining:

> Claims to have 1TB but only has 931GB as reported by my operating system.

The "missing" 69GB isn't due to formatting or any misdoing on the part of the drive manufacturer, it's because the OS is using the wrong units.

> Even if we did use the kilo vs. kibi terms, HDD and SSD manufacturers are the ones who are getting it right.

I'd remove that "even".

If we used both terms, then they'd be getting it right.

But since we're not, they're being misleading compared to the typical use.

> I think it's the memory people who got this wrong

The memory people? It's pretty much everyone that uses "gigabyte" and isn't selling you a storage device.

Many Linux utilities do things with the kibi terms, using the MiB, GiB, TiB units.

There have been a few lawsuits about this: https://en.wikipedia.org/wiki/Binary_prefix#Consumer_confusi...

The outcome is that hard drive manufacturers need to explicitly print their definition of a "gigabyte". IMHO, it's not a right or wrong discussion, just a difference.

The age old usable capacity complaint. (i.e. a 1TB drive will never actually give you 1TB of usable space)

Some people use decimal prefixes but they mean the binary values, for example kilo is 1000 but in computers is most often used as 1024. HDD manufacturers express the size in decimal (1 TB = 10^12 bytes), people that think in 1024 multiples will calculate this is less Gigabytes than expected.

1GB RAM > 1GB HDD

I never understood why specifically HDD manufacturers are criticized when in fact all vendors and all types of hardware (expect RAM) follow the same convention of using decimal prefixes: http://blog.zorinaq.com/decimal-prefixes-are-more-common-tha...

I see this repeated time and time again, and every time it isn't true.

This is what my console shows when it sees a 1 TB drive:

[ 1.629496] sd 2:0:0:0: [sda] 1953525168 512-byte logical blocks: (1.00 TB/932 GiB)

If you multiply those 2 numbers (1953525168 x 512 bytes), you arrive at 1000204886000 bytes, or 1000.204886 GB. You can write to every single one of those bytes/sectors, none of them are reserved.

Your choice of filesystem dictates the maximum amount of space you can use on the drive, but the drive itself is giving you all of the capacity you asked for, and if you don't use any filesystem at all, you can write 1TB to it.

The confusion comes from Microsoft Windows being involved. Windows measures in IEC units, but displays the value as an SI unit.

So, for example, it will tell you that a 16000000-byte file is 15.26 MB. It isn't. It's 16 MB; it is also 15.26 MiB. Likewise, Windows will tell you that a 1 TB harddrive is 931 GB. It isn't, it's 1000 GB; it is also 931 GiB.

If I'm a cable vendor, and you ask me for a reel of 8 kilometres of cable, and I give you a 5-mile reel, I gave you what you asked for. 5 is less than 8, but 5 miles is not less than 8 kilometres. 931 is less than 1000, but 931 GiB is not less than 1000 GB.

> I see this repeated time and time again, and every time it isn't true.

What isn't true? That they're using a definition that's technically correct but doesn't align with the lay definitions of the units?

I don't think you can call that untrue, no matter how prescriptivist you get.

And it's worse than that, because other technical users also mean powers of 2^10.

And it goes far beyond Windows, and has nothing to do with file systems.

Gibibyte is not a different definition of gigabyte. They’re different units.

On reading this on iOS I selected the word and pressed “look up”. It seems that iOS strips out formatting in its dictionary, make superscript full size. Not ideal. Gigabyte “a unit of information equal to one thousand million (109) or, strictly, 230 bytes.”

When they shorten it to “G” they are absolutely trying to deceive.

I though it was SI retroactively standardizing the Ki Mi Gi prefixes in an attempt to fix things, and before that the common usage of GB was just 2^30 Bytes.

The SI system can give a definition of GB, but that doesn't mean they have given the definition of GB.

Neither. G is the SI suffix Giga, and always means 1,000,000,000. GiB is an IEC unit, not an SI unit. They're from entirely different bodies.

That's exactly what he meant.

Technically, everyone in computing BUT the HDD manufacturers are doing this. That's why we now have MiB, GiB etc. units as well as MB, GB etc.

Reply on news.ycombinator.com