PowerPill Question

Horris · 2010-02-08 23:05:42

Hi People,

Why the difference of size for same package?

 sudo pacman -S chromium
resolving dependencies...                
looking for inter-conflicts...           

Targets (1): chromium-4.0.249.78-3  

Total Download Size:    17.41 MB
Total Installed Size:   46.11 MB

Proceed with installation? [Y/n] n

$ sudo powerpill -S chromium

:: The following packages will be downloaded:
   [extra]
     chromium

:: Number of Packages:               1
:: Total Download Size:       18.26 MB
:: Total Installed Size:      48.35 MB

Last edited by Horris (2010-02-08 23:06:10)

skottish · 2010-02-09 01:07:46

Are you sure that they are the same package? It may be that one repo has a newer release than the other.

Xyne · 2010-02-09 01:18:43

Pacman's output is incorrect. It's actually reporting the sizes in MiB (Mebibytes) instead of MB (Megabytes).

The size of the chromium package is 18256138 B (Bytes)
18256138 / (1024^2) ≈ 17.41 MiB
18256138 / (1000^2) ≈ 18.26 MB

Open a ticket about this on the bug-tracker.

Last edited by Xyne (2010-02-09 01:18:59)

Aidenn · 2010-02-09 10:47:19

I'm not sure it's a bug. using MB for Mebibytes has been accepted for decades. sure, it's incorrect, but it's how everything hardware-related is labeled and counted (besides the stickers on hard drives ;p). If a change should be made, it's the one of renaming MB to MiB in programs, not making them count in powers of ten.

Ultraman · 2010-02-09 11:53:58

Aidenn wrote:

I'm not sure it's a bug. using MB for Mebibytes has been accepted for decades. sure, it's incorrect, but it's how everything hardware-related is labeled and counted (besides the stickers on hard drives ;p). If a change should be made, it's the one of renaming MB to MiB in programs, not making them count in powers of ten.

Accepted... Accepted... No, more like forced down your throat. Like you said: Hard drive companies do this.
Why is pacman doing it? I don't know. But it's not proper behaviour and deserves a bug report imo.

Last edited by Ultraman (2010-02-09 11:54:28)

Aidenn · 2010-02-09 13:43:18

Ultraman wrote:

Accepted... Accepted... No, more like forced down your throat. Like you said: Hard drive companies do this.

nope, hard drive companies got it correctly. ;)

they say 20 GB, which means 20 Gigabytes or 20*1000^3 Bytes. it translates to 18.63 GiB, or Gibibytes.

Ultraman wrote:

Why is pacman doing it? I don't know. But it's not proper behaviour and deserves a bug report imo.

pacman isn't doing it, powerpill is. it's correct, it says MB and it is a Megabyte (1000^2). it's the same convention hard drive companies use.

however we, as in computer science geeks or programmers, operate on Gibibytes, yet we write GB and say "gigabyte", because that's how it was for decades. pacman is correct with its counting (a power of two, just how we like it), but writes "MB" because of the tradition. the only change it may need is changing the string "MB" to "MiB", not changing from power of two to power of ten.

Last edited by Aidenn (2010-02-09 13:59:41)

jowilkin · 2010-02-09 14:19:11

Aidenn wrote:

I'm not sure it's a bug. using MB for Mebibytes has been accepted for decades. sure, it's incorrect, but it's how everything hardware-related is labeled and counted (besides the stickers on hard drives ;p). If a change should be made, it's the one of renaming MB to MiB in programs, not making them count in powers of ten.

The prefix "mebi" came into use precisely because of this problem. Before it was used the abbreviation MB was ambiguous. Now we have precise notation, let's use it for God's sake!

Just as Xyne pointed out, Pacman should report 17.41 MiB or 18.26 MB. Reporting 17.41 MB is incorrect.

xduugu · 2010-02-09 15:42:05

Practically no program uses MB for 10^6 bytes (apart from os x 10.6), thus powerpill's behavior is imo way more confusing than the MB vs. MiB thing. Actually, I do not remember a single program I use that uses MiB (ls, df, du etc. use 1024 as standard configuration) and therefore I usually interpret MB as MiB.

Xyne · 2010-02-09 17:21:18

Just because something has been done the wrong way for decades doesn't mean that it should be accepted. That's the very attitude that makes it so difficult to implement standards and get things fixed when they're actually broken. It's also related to the "good enough" mentality which impedes progress in general.

I could easily switch Powerpill to "MiB" (my size function accepts the size as a parameter). It's not incorrect in either case though, although Pacman is. It should replace "MB" with "MiB" in the output, which is probably simpler to implement, or it should switch to using powers of 1000.

xduugu · 2010-02-09 18:16:48

I think you didn't get my point. Even if it's technically correct to use MB for 10^6 bytes (at least since 1999), you cannot simply use the term "MB" for something different than everybody else does without confusing users. I did not say that pacman should continue to use "MB", just that using the decimal prefix is (currently) a bad idea in my opinion.

Aidenn · 2010-02-09 20:29:46

Xyne wrote:

Just because something has been done the wrong way for decades doesn't mean that it should be accepted. That's the very attitude that makes it so difficult to implement standards and get things fixed when they're actually broken. It's also related to the "good enough" mentality which impedes progress in general.

actually, it is implemented as a standard. IEEE states that using mega- giga- etc. for powers of two in computing is an allowed exception. ;)

Xyne wrote:

I could easily switch Powerpill to "MiB" (my size function accepts the size as a parameter). It's not incorrect in either case though, although Pacman is. It should replace "MB" with "MiB" in the output, which is probably simpler to implement, or it should switch to using powers of 1000.

I think using power of ten, even while using the correct unit multiple name, is more inconvenient than using power of two with incorrect unit name. I do agree that using MiB and GiB is better and should be the way, but I don't think using MB or GB for power of two values is a bad thing. I think that making programs that say "MB" and use power of ten is disorienting, even if correct. that's because there's no point in using real megabytes or gigabytes in computing, it's a redundant unit.

to summarise my point of view:

a) MB as mebibytes is semi-incorrect, but everyone knows what it means anyway
b) MiB as mebibytes is correct and the best way to do it
c) MB as megabytes is correct, but shouldn't be used at all, because it's redundant and 99,9% of people will assume it means mebibytes anyway. that's why we hate HD manufacturers ;)

Last edited by Aidenn (2010-02-09 20:34:42)

Xyne · 2010-02-09 21:47:12

Powerpill and Bauerbill now use MiB:

:: Number of Packages:                       1
:: Total Download Size:              17.41 MiB
:: Binary Package Install Size:      46.11 MiB

I still think Pacman should display "MiB", but I'm not really bothered. I'll leave that quest to the White Knights of the Order of Standards.

I still think it's ridiculous to ignore standards because "that's just what way it's always been done" (it's not a standard if it has such prominent exceptions). It's just plain laziness.
Example: http://en.wikipedia.org/wiki/File:Metri … on_map.svg

xduugu · 2010-02-09 23:10:54

Nobody is ignoring standards. Isn't "MiB" a standard, too?
Anyway, the SI prefixes just does not make sense currently, because the majority of software use them in a wrong manner. And therefore, the correct use of SI prefixes creates confusion which is the exact opposite of what a standard is intended for.

Xyne · 2010-02-09 23:44:04

xduugu wrote:

Nobody is ignoring standards. Isn't "MiB" a standard, too?
Anyway, the SI prefixes just does not make sense currently, because the majority of software use them in a wrong manner. And therefore, the correct use of SI prefixes creates confusion which is the exact opposite of what a standard is intended for.

So basically we should all follow IE browser conventions when deciding on HTML standards, or any of Microsoft's other de facto standards because they control the majority of the market?

Please tell me that you see the problem with this line of thinking.

Aidenn · 2010-02-09 23:55:47

Xyne wrote:

So basically we should all follow IE browser conventions when deciding on HTML standards, or any of Microsoft's other de facto standards because they control the majority of the market?
Please tell me that you see the problem with this line of thinking.

yes, using MiB is the preferred way. but MB for mebibytes is fine too.

and to answer your question - no, because no standards were created based on Microsoft's semi-conscious blabbering. there are HTML standards, to which Microsoft doesn't conform, but we shouldn't care about it at all. they were there before Microsoft started fiddling with the WWW*. with MiB/MB confusion it's all different, because back when kB was first used as a definition for 1024 bytes, there was no kibibyte. it was simply the closest thing and "good enough". kibibyte was created only 10 years ago, which means we are still in the transition period. eventually, everyone will/should be using MiB and KiB, but for now still using old terms isn't a faux pas.

* IE8 is actually a nice piece of GUI. if only it would conform...

Last edited by Aidenn (2010-02-10 00:03:50)

xduugu · 2010-02-09 23:56:19

I think your comparison is flawed, because the defacto standard would be using MB for MiB.

jac · 2010-02-10 00:03:53

If we're in a transition period, things should be transitioning. Preferably towards the correct way of doing things, i.e., using the *bibyte units.

Xyne · 2010-02-10 00:10:38

xduugu wrote:

I think your comparison is flawed, because the defacto standard would be using MB for MiB.

The scientific usage of kilo, mega, giga etc precedes computer usage. The already were de facto standards. Early programmers misused them and set a precedent that we're still following. It's a convention, not a standard as I see it. We not have *bibytes, which we should all start using when using binary powers.

Aidenn · 2010-02-10 00:12:36

jac wrote:

If we're in a transition period, things should be transitioning. Preferably towards the correct way of doing things, i.e., using the *bibyte units.

yeah, that's why Xyne made PowerPill the way he did (he just didn't use the most popular unit, but corrected it in this very topic). Pacman too should be saying MiB. and other applications. but if you don't change it, it's still not a bug, there's no deadline for the transition. the standard says that using old, incorrect prefixes is allowed. and if the standard says that - it can't be a bug. it depends wholly on programmer's attitude. ideally everyone should be using the new prefixes, but it's not set in stone.

Xyne · 2010-02-10 00:24:54

Aidenn wrote:

yeah, that's why Xyne made PowerPill the way he did (he just didn't use the most popular unit, but corrected it in this very topic). Pacman too should be saying MiB. and other applications. but if you don't change it, it's still not a bug, there's no deadline for the transition. the standard says that using old, incorrect prefixes is allowed. and if the standard says that - it can't be a bug. it depends wholly on programmer's attitude. ideally everyone should be using the new prefixes, but it's not set in stone.

No no no. We need large, angry mobs with weapons. We need to get the prefix revolution going. Death to all i-less binary prefixes and the scum that refuse to change them!

Bring some food too... this may take a while... and ffs, don't tell RMS about this, or else we'll have people calling them GNU/Mebibytes.

Aidenn · 2010-02-10 00:28:30

Xyne wrote:

Bring some food too... this may take a while... and ffs, don't tell RMS about this, or else we'll have people calling them GNU/Mebibytes.

That just made my day. And the M stands for Mebi'snotmega.

Xyne · 2010-02-10 00:36:28

Aidenn wrote:

That just made my day. And the M stands for Mebi'snotmega.

xduugu · 2010-02-10 00:36:47

Xyne wrote:

The scientific usage of kilo, mega, giga etc precedes computer usage. The already were de facto standards. Early programmers misused them and set a precedent that we're still following. It's a convention, not a standard as I see it. We not have *bibytes, which we should all start using when using binary powers.

I know they misused it, but I'm pretty sure there would be another binary based unit if they hadn't. However, it would be funny to buy 4.1 GB RAM modules.
Anyway, I don't see why it's such a big issue to use *iBytes. If it's just to follow a standard, you have to call it Megaoctets anyway, because there is no standard that says a byte consists of 8 bit.

To summarize and hopefully end this discussion: I think you should use *iBytes and avoid the si prefixes (at least until the majority switched to the correct usage of si prefixes / *iBytes).

Peasantoid · 2010-02-10 01:16:46

Xyne wrote:

http://en.wikipedia.org/wiki/File:Metri … on_map.svg

Ayup. This is one of the annoying things about living in the US.

ngoonee · 2010-02-10 01:51:18

Xyne wrote:

Aidenn wrote:
yeah, that's why Xyne made PowerPill the way he did (he just didn't use the most popular unit, but corrected it in this very topic). Pacman too should be saying MiB. and other applications. but if you don't change it, it's still not a bug, there's no deadline for the transition. the standard says that using old, incorrect prefixes is allowed. and if the standard says that - it can't be a bug. it depends wholly on programmer's attitude. ideally everyone should be using the new prefixes, but it's not set in stone.
No no no. We need large, angry mobs with weapons. We need to get the prefix revolution going. Death to all i-less binary prefixes and the scum that refuse to change them!
Bring some food too... this may take a while... and ffs, don't tell RMS about this, or else we'll have people calling them GNU/Mebibytes.

That so brightened up my morning. Thanks Xyne!

Peasantoid wrote:

Xyne wrote:
http://en.wikipedia.org/wiki/File:Metri … on_map.svg
Ayup. This is one of the annoying things about living in the US.

You mean besides the 'living in the US' part? =p

Arch Linux

#1 2010-02-08 23:05:42

PowerPill Question

#2 2010-02-09 01:07:46

Re: PowerPill Question

#3 2010-02-09 01:18:43

Re: PowerPill Question

#4 2010-02-09 10:47:19

Re: PowerPill Question

#5 2010-02-09 11:53:58

Re: PowerPill Question

#6 2010-02-09 13:43:18

Re: PowerPill Question

#7 2010-02-09 14:19:11

Re: PowerPill Question

#8 2010-02-09 15:42:05

Re: PowerPill Question

#9 2010-02-09 17:21:18

Re: PowerPill Question

#10 2010-02-09 18:16:48

Re: PowerPill Question

#11 2010-02-09 20:29:46

Re: PowerPill Question

#12 2010-02-09 21:47:12

Re: PowerPill Question

#13 2010-02-09 23:10:54

Re: PowerPill Question

#14 2010-02-09 23:44:04

Re: PowerPill Question

#15 2010-02-09 23:55:47

Re: PowerPill Question

#16 2010-02-09 23:56:19

Re: PowerPill Question

#17 2010-02-10 00:03:53

Re: PowerPill Question

#18 2010-02-10 00:10:38

Re: PowerPill Question

#19 2010-02-10 00:12:36

Re: PowerPill Question

#20 2010-02-10 00:24:54

Re: PowerPill Question

#21 2010-02-10 00:28:30

Re: PowerPill Question

#22 2010-02-10 00:36:28

Re: PowerPill Question

#23 2010-02-10 00:36:47

Re: PowerPill Question

#24 2010-02-10 01:16:46

Re: PowerPill Question

#25 2010-02-10 01:51:18

Re: PowerPill Question

Board footer