| Date: | October 18, 2006 / year-entry #352 |
| Tags: | other |
| Orig Link: | https://blogs.msdn.microsoft.com/oldnewthing/20061018-03/?p=29343 |
| Comments: | 61 |
| Summary: | Help text is not the place to put logic puzzles. In Windows Vista, when you go to the System control panel, you are shown a number that describes your computer's rating. But are higher numbers better or worse? If I had a choice, would it be better to have a 1 rating or a 5 rating? In earlier... |
Help text is not the place to put logic puzzles. In Windows Vista, when you go to the System control panel, you are shown a number that describes your computer's rating. But are higher numbers better or worse? If I had a choice, would it be better to have a 1 rating or a 5 rating? In earlier betas of Windows Vista, you had to have a degree in philosophy to figure this out. If you clicked the "Help" button on the System control panel page, you were sent to a page of help text that tried to explain the performance rating. When it got around to explaining what the number means, the text said, paraphrased, "When looking for software to run on this computer, you should choose programs whose rating is less than or equal to the rating of this computer." So does this mean that bigger ratings are better? "Well, if a program's rating is small, then the computer's rating needs to be bigger than that, so a program wants its rating to be as small as possible so more computers can run it. If my computer's rating is small, programs will be fighting to get a rating low enough that I can run it. That's a good thing for me, right? No wait, but what if the program I want has a high rating? Then my computer will need a higher rating. If my computer had a low rating, then that wouldn't be less than or equal to the program's rating. No wait, I got it backwards, it's the program that needs to be less than or equal to the computer, not the other way around. If the program's rating needs to be less than or equal to the computer's rating, then that means that the computer's rating needs to be greater than or equal to the program's rating. If my computer rating were higher, than it could run more programs." "I think." "Why can't they just say, 'Bigger numbers are better'?" |
Comments (61)
Comments are closed. |
Raymond Chen at his best : "Help text is not the place to put logic puzzles."
"Why can’t they just say, ‘Bigger numbers are better’?"
Even better:
Computers with bigger numbers have better performance. Programs will be marked with a performance requirement your computer will have to meet, if you want to use that software efficiently.
And what happens if the software has different performance markers? For example, turn off some of the eye candy and the rating goes down?
Could get very confusing!
@koehli: "performance"? "requirement"? Are you trying to design the packaging for Linux distro?
Amusing dicotomy. If you don’t want people to understand that bigger is better, why do the rating in the first place?
I think this rating thing is a big mistake, but I’m sure the forces for it have compelling arguments.
@Gary Williams
Basically yes. You could have a slider like this
—- Your current rating is: X.XX —-
|—[o]——————————-|
make me pretty make me fast
@kiwiblue
Actually I was using the word requirement in it’s everyday meaning and not in it’s techno-geek meaning. So companies can put a sticker on their hardtop packages. Like:
"This software requires one of the operating system Win98/98/ME, NT4.0, Win2000, WinXP/2003"
"This software requires Windows Vista with a performance rating of at least 4.5"
I must say, I don’t see any confusion here.
The help text seems clear to me… higher is better. The higher my computer’s number (performance), the more software it can run.
@koehli:
"Programs will be marked with a performance requirement your computer will have to meet, if you want to use that software efficiently."
Do I need to take my computer somewhere, so it can meet the performance requirement? Do I need to call first and make an appointment?
I wonder if a slightly more complicated numbering scheme would be more serviceable. IE: GraphicsProcessorMemory and maybe Storage. Each item can have its own rating scheme say from 1 to 5. (Note: firefox wont let you type a forward slash in a textbox, thus the backslash separation :))
So your computer can have a GPMS rating of 1335. That would mean that it probably needs a G upgrade. With a single number it is much harder to explain what it means, to manufacturers and to users because it must have some sort of non linear scaling factor built into it.
@kiwiblue:
"Do I need to take my computer somewhere, so it can meet the performance requirement? Do I need to call first and make an appointment?"
Well, first, you find a nice software package for your computer on Craigslist and setup a candlelight dinner by the bay…
The reason higher numbers mean bigger performance is so that it can be an open-ended scale. Why do many people think the score is capped at 5? If that was true it would be a mess. This seems like one of those Raymond "you already know the answer" or "imagine if it worked that way" kind of situations.
Five years from now, given hardware performance advances, everyone would be clustered between 4.9 and 5 with no differentiation among fast and (comparatively) slow hardware. You’d need to "recalibrate" the scale occasionally and then people would start comparing scores that used different calibrations. You’d see comments like "I’m a 3.9 using Vista Gold, but it dropped to 3.2 after the Vista SP1 recalibration. So you’re running SP2 and have a 2.9? What would that be in the SP1 recalibration?"
A slightly more complicated numbering scheme? It’s appealing to the geek inside me, but the average user won’t understand why their 1343 program won’t run on their 4225 computer. After all, 4225 is a larger number than 1343.
Stu, the system rating tool (WinSAT) already runs on XP because it’s part of the Vista Upgrade Advisor. It’s a command line tool and I believe it doesn’t have a lot of dependencies so you should just be able to copy winsat.exe to XP and get some numbers. Microsoft would need to clarify whether it can be distributed with third-party software as a rating tool for XP.
Graphics/Processor/Memory
Works for me in Firefox
At the company that I am at currently they have a 1-5 scale for internal ratings of things that goes something like this.
1 – Adequate
2 – Needs Work but will suffice
3 – Not Adequate but we might accept in a pinch
4 – Awesome
5 – Absolutely horrible
Now THAT is useless.
@koehli
and how do you put that on a box?
"And this package requires a vista performance rating of 4.5 unless you turn off the eye candy then it’s 2.0 unless it’s a thursday and you want to run the maths module with high precision then it’s 3.2 or if you put on the first two bits of eye candy then it’s 3.193", etc, etc!
Also, What does Vista base this rating on? does it do any checks? You could easily have two machines of the same make side by side with different ratings because one is using a 5,400rpm HDD and one a 7,200rpm HDD….
@kiwiblue:
"Do I need to take my computer somewhere, so it can meet the performance requirement? Do I need to call first and make an appointment?"
Seems to be common dating practice on the internet ;-)
Google: Results about 55,200,000 for "meet a requirement".
PS: As for the Latin Lover qualities of the Anglophone web: "meet a requirement" beats "make an appointment" by a sound 10%.
Google: Results about 50,700,000 for "make an appointment"
The other problem is that I cannot find a max value anywhere. So if I get 4.5, is this good or bad?
Is 4.5 out of 5 (sweet!), or 4.5 out of 10 (bummer)?
were a max value. What happens if the max is 10 and you buy a 10
computer, and then an even faster computer comes out next year – what
rating does that computer yet? -Raymond]
@ Raymond’s interjection
True, but IIRC doesn’t it say somewhere that the scores are relative to each other. Talk about confusing. I have a 1.8 Ghz Turion 64, with 1 gig of DDR ram and an ATI Radeon Xpress 200M. (128 dedicated DDR Ram) I score great with Ram (4.0) and processor (3.9) But because of my "pitiful" graphics card relative to the rest of my system I’ve got a 2.4 overall rating. Now wait a second, my system is pretty dang smoking for a laptop, not as good as a desktop yes, but probably better than a 2.4 especially since technicnally I meet the requirements laid out for "Vista Premium Capable" or whatever they call the level above "capable."
Furthermore, imagine the poor user here.
"Hmm My ram is the choke point, I’ll go buy more."
So they pay Big Box Electronics store to put in more Ram, this causes the scores to readjust so that they actually go down! (It’s possible.) I mean Yikes!
@Gary Williams
No, the box just demands what it needs (Your computer must have a performance rating of 5.0 or more). The "Computer"(the OS in reality) can either be tuned with a make me pretty/ make me fast slider to be fast or pretty, or just come at a default rating.
The user interface seems to be pretty clear, question remains if the multivariate entity "performance" can be reduced to a scalar classifier in any meaningful way. Put differently:
Is there a strong covariance for commonly used computer configurations between the most important single performance indicators like HD-speed, HD-capacity, Processor speed, graphics card power, etc?
My guess would be that MS has done some statistical analysis over the data they got from their "customer feedback programs" and the answer is yes and also the data has been used in a cluster analysis to produce the "performance rating categories" used in Vista.
greets, koehli
PingBack from http://blog.labnotes.org/2006/10/18/rounded-corners-49/
"I wonder if a slightly more complicated numbering scheme would be more serviceable. IE: GraphicsProcessorMemory and maybe Storage." –James
Please be careful with this. You are getting uncomfortably close to educating the users.
-Wang-Lo.
[Imagine what the world would be like if there were a max value. What happens if the max is 10 and you buy a 10 computer, and then an even faster computer comes out next year – what rating does that computer yet? -Raymond]
Duh, you turn it up to 11! (Rawk out.)
After all, 11 is louder than 10.
"Is there a strong covariance for commonly used computer configurations between the most important single performance indicators [SPIs] like HD-speed, HD-capacity, Processor speed, graphics card power, etc?" –koehli
Yes, certainly. Just visit a computer retail outlet or examine a manufacturer’s catalog.
Personal computers are a commodity product. Margins are small. Manufacturing, stocking, and especially facing a wider selection of independent combinations of SPIs would raise prices across the board.
I often purchase PCs for my clients. I spec out the exact SPIs required, then comparison-shop. Invariably, the cheapest retail catalog selection that meets or exceeds each of my criteria is cheaper than a whitebox locally made to my exact specs.
-Wang-Lo.
I could see this being a major problem.
Say you have a great app, it’s rating may be a 1(one) because it knows how to use the power properly and runs awesome. Then you could have some bloatware that rates a 5(five). Which one do you think most consumers will buy?
The bloatware of course because it requires a beefier machine which means it must be better!
OMFG, I can predict the future!
At 11:26, I post a "imagine" scenario about there being no maximum score.
At 12:21, Mihai asks about the maximum score, and Raymond replies with an "imagine" scenario.
Bow before me!
Oh yeah, the Tigers will win the World Series.
I think the tech writers try to avoid value judgements. So they can’t just say "Higher is better" because it might not be better for everyone. Nor can they suggest that "Higher lets you run with better performance" because it might not be true.
So instead they stick to a factual statement: pick programs that are able to run on your computer.
Hm…
Here are my scores:
4.8 – Overall
4.8 – Processor
4.9 – RAM
5.9 – Graphics
5.7 – Gaming graphics
5.2 – Primary hard disk
Notice that most of my scores are over 5. Also, notice that the "overall" score is really just the lowest of the individual scores. Nothing is "relative" or on a scale from 1 to 5. All of these ratings are absolute, and reflect real performance metrics such as frames per second on the GPU, calculations per second on the CPU, and data transfer rate on the hard disk.
Well, at least it’s a step up from GetSystemMetrics(SM_SLOWMACHINE).
So, you’ve implemented "system too slow, come back when you’re ready to play"?
I got confused trying to understand your confusion.
It seems to me that Microsoft assumes that the moment Vista is released, everybody will switch to it, overnight.
In reality, Windows XP will probably still be the most pervielent OS in 2 years time, and some businesses will probably still be using it in 5 years time.
Therefore, unless Microsoft decides to backport the system rating tool, software houses who still want to support XP (almost all of them) will still have to put system requirements on the packaging, instead of relying solely on the rating system.
So the rating system just adds *another* piece of information that must be printed on the packaging, rather than simplifying anything.
Plus the almost certianty that PC OEM’s will work out ways to "cheat" the rating system and the uncerianty of what happens with Vista+1 (Will it use the same numbering? Will we end up with PC’s rated in the hundreds?), this admirable attempt at simplifacation seems more like added complexity…
I don’t like reading technical stuff and having to understand it before I make a decision. ‘Bigger numbers are better’ would work for me.
I’m a game developer, so this performance rating stuff is a pretty big deal to me. I’m not entirely convinced that this is going to be the solution to all our problems, but it’s worth a shot.
Consumers are simple people, and marketing people especially so. For example, have you noticed how games still include things like "64MB video card required" on their box? This despite the fact that the amount of video memory hasn’t been the most salient feature for determining compatibility for many years. It’s just that marketing has decided that it’s the number that consumers understand so that’s what they go with. (I’m not sure why they think that, since the average consumer is barely aware of which operating system they’re running, much less how much video memory they have.)
So what the box actually says is something more like "16MB/3D DirectX 9.0 or later*", which isn’t sufficient either, but there’s some really fine print crammed in listing a bunch of the more common cards which actually work and note to the effect that other stuff may work too but don’t count on it.
It would be really nice to make this whole problem just go away. Boiling your system performance down to a single number that a consumer can actually understand would be great…if it can be made to work.
I don’t think that continuous ratings are going to work very well, due to the good-enough problem. If I have a similar computer to my neighbor, and he gets a 5.0 while I get a 4.9, what does that mean? Can I still run 5.0-rated programs? I think most people will just say that 4.9 is good enough and try the program anyway. But then you have problems with people who try it with a 4.8 machine, or 4.5….
With tiers, you can eliminate some of this ambiguity. A graphics card, theoretically, is either shader model 3.0 or it isn’t; cards don’t show up as supporting 2.9 in a config dump. Unfortunately, trying to establish tiers across many different components with many IHVs with very creative marketing departments is difficult.
Jonathan,
The google earth approach, as you call it, is a lot of noise that signifies nothing. If it was really of any use, why do they still need to print a ‘traditional’ spec below it?
The reason the Google Earth specs work is because I can go buy the cheapest Dell they sell and Google Earth will work on it.
However, I can’t buy the cheapest Dell or even a lot of mid-range Dells and expect /insert newest high-end graphics game/ to work with it.
However, I could have a two-year-old computer that runs /insert newest high-end graphics game/ perfectly if it were crafted properly.
Why still have traditional spec listing: Because geeks like those types of things.
The Google Earth approach :
Google Earth is a broadband, 3D application that not all computers can run.
* Desktop computers older than 4 years old may not be able to run it.
* Notebook computers older than 2 years old may not be able to run it.
(http://earth.google.com/download-earth.html)
They do ave the traditional requirement spec below, though.
Only one type of consumer level application has trouble running on most machines: Games.
The performance rating system needs to address gaming specifically and needs to be branded and marketed as such, ie. Game Peformance Score.
This should remove some ambiguity and for people not interested in games, it would allow them to ignore this number.
The obvious question is, who decides the numbers for each piece of software? The developers themselves? Microsoft?
Would a developer release a piece of software and give it a high rating, effectively telling people with slower machines to not buy it, or give it a low rating in order to generate more sales, regardless of performance?
I love how AMD x2 4800+ at 2.4GHz has lower rating than my Core 2 Duo E6300 at 1.86GHz. I bet all AMD fans will love it too ;)
Tim, I’ve wondered… exactly how slow does the machine have to be before GetSystemMetrics(SM_SLOWMACHINE) tells you it is slow?
Nick: "Would a developer release a piece of software and give it a high rating, effectively telling people with slower machines to not buy it, or give it a low rating in order to generate more sales, regardless of performance?"
Given the tendencies of the software industry so far, I’m picking the latter. Microsoft like to pretend that you can run Windows XP on a 300MHz machine with 128MB RAM, for example.
Some game publishers are the masters at this game though.
James: "Only one type of consumer level application has trouble running on most machines: Games."
I thought the performance rating thing was meant to help people figure out whether they could run Qua… sorry, Aero Glass. It’s demanding in a similar way to games (although obviously rather less so than most modern games).
Games, in particular, have specific graphics card requirements. If you want a high frame rate in Quake or Doom, you need a mucho-high-end graphics card. Sheesh, I have seen graphics cards that list for $2500.
Since I never need a high frame rate, as I am always working in Excel or Visual Studio .NET or SQL 2005 Management Studio or Outlook or looking at pages in Google, I am content with built-in graphics. (One of my office computers has built-in graphics, and the other one has a low-end card.)
I would hope that using built-in graphics would not pull down my overall score, but I’ll bet it does.
Thursday, October 19, 2006 4:19 AM by Jonathan
Oh no. Google Earth is going to stop working on my newest notebook … um, a few days ago, I think. I’d better go check it ^_^
Thursday, October 19, 2006 3:00 PM by Igor
But are you running 32-bit Windows or x64? The shortage of drivers for x64 might push down the number on that.
"I love how AMD x2 4800+ at 2.4GHz has lower rating than my Core 2 Duo E6300 at 1.86GHz. I bet all AMD fans will love it too ;)"
Igor, looks like you are still conditioned by the mega(Giga?)hertz war of yesteryear. There are many detailed articles describing the Core architecture and comparing it to the current AMD offering. Most agree that the Core architecture has many enchancements that make it a better performer than the AMD Hammer architecture even at lower clock speed. It is then definitely possible that a lower clocked processor will have a real performance advantage over its higher clocked market counterpart. Hence the performance number advantage. No surprises there then.
IMO: ditch numbers for WinSAT, or make them smaller and put them in parenthesis off to the side.
Use words instead:
– Poor
– Fair
– Good
– Excellent
etc. Much easier to figure out if a number is good or not.
Also it’s not a scale from 1 to 5… one of my components scored a 6 (I forget which, I’m not in Vista at the moment).
>But are you running 32-bit Windows or x64?
I tried with 32-bit Vista RC2.
>The shortage of drivers for x64 might push
>down the number on that.
For my hardware all x64 drivers are available already so that doesn’t hold water.
>Igor, looks like you are still conditioned
>by the mega(Giga?)hertz war of yesteryear.
No I am not. I was just being sarcastic :)
@Raymond:
IMO, Microsoft already made a mistake here. The most powerfull
systems of today should score 1.0 so that there is a room for
improvement.
I also agree that the descriptive labels would be much more appropriate here. Why?
First because there is no need to up the hostility between people by
allowing them to compare performance numbers and say “My rig is 7.43
times faster than yours”.
Second, any hardware which is rated “Poor” will perform poorly and
any new hardware which exceeds “Excellent” will still perform excellent
Actually, any advantage of its better performance will not be
observable at least when it comes to the operating system itself.
Third, having “Poor” in bold face next to your graphics card icon is
a clear signal what do you need to upgrade. On the other hand, having a
number between 1.0 and 1.9 is not as clear especially to those who
don’t use Arabic numbers.
Fourth, number range between 1.0 and 1.9 is then simply a stretch of
a word “Poor” into “Poorest”, “Poor”, and “Less poor” — i.e.
completely useless subdivision.
Also, the scoring scheme is fundamentally flawed with the overall score being equal to the lowest one.
If an Average Joe has overall of 4.8 determined by his current
dual-core CPU score, and he plugs quad-core CPU in expecting to see a
larger overall number he will be greatly disappointed to see that he
has overall of 5.2 determined by his HDD even though his CPU has
blazing 9.6.
instead of numbers please also include how they anticipate the ratings
for the next version of Windows to work? Do you want “Excellent” to be
redefined? (So a “Windows Vista Excellent” is equal to a “Windows
Vista+1 Medium”.) Or do you want to keep the name the same and just
invent new power levels? (So Windows Vista+1 might have a new “Awesome”
level that is a level above “Excellent”.) I already described the
drawbacks of each of these systems. -Raymond]
> For my hardware all x64 drivers are available already so that doesn’t hold water.
But you were using the 32-bit OS, so it *does* actually hold water. (I do not believe the argument was "32-bit drivers slow down a 64-bit OS", because AFAIK you can’t *run* 32-bit drivers on a 64-bit kernel. I believe the argument was "people don’t run a 64-bit OS because their hardware doesn’t have 64-bit drivers, and this is slower". You were running a 32-bit OS.)
>Could all the people who advocate using words
>instead of numbers please also include how
>they anticipate the ratings for the next
>version of Windows to work?
Raymond, I am not arguing which of the two schemes is more future proof because truth to be told neither is. I just claim that words are much easier to understand and I believe they create less confusion.
I guess you are well aware that numbering also won’t work.
If Vista+1 comes out in say 5 years, all 5 year old hardware which had “Excellent” rating on Vista+0 will most likely have “Poor” rating on the new one.
If you call that redefining words, how different is that from redefining numbers?
You will either have to:
a) use the same range of numbers where Vista+0 5.0 will be redefined to Vista+1 1.0.
or:
b) make it even more confusing by using a new range of numbers (say 5.0 to 10.0) for Vista+1.
>Do you want “Excellent” to be redefined?
There is really no need for that. “Excellent” in Vista+1 is also “Excellent” in Vista+0. On the other hand “Poor” in Vista+1 is still “Excellent” in Vista+0 while “Poor” in Vista+0 most likely won’t meet minimum hardware requirements for Vista+1. I really do not see a problem there.
What I want to point out is that there is really no need to have continuity here. It is not feasible to have it either. Why? Because hardware changes a lot in a short time span.
You get new CPU instructions, new DirectX version along with a host of new hardware accelerated rendering features. Old OS (Vista+0) will not be able to take advantage of those new features and as a result any performance comparison with Vista+1 will be invalid because you are not comparing apples to apples.
Anyway, I believe that this measure of performance be it words or numbers should not be comparative.
It should be tied to the current OS release because it gauges user experience on particular hardware in the current OS.
As always, if you have a different picture from your corner, please enlighten us.
Everyone sorry for O/T.
@BryanK:
Man, I said I was running 32-bit Vista and I compared my result with Athlon x2 4800+ which was also running 32-bit Vista.
If you are implying that my setup will run slower in 64-bit Vista because of lack of 64-bit drivers then I am calling BS because I have the drivers, I just haven’t bothered to download the 64-bit OS itself because it is another 2.5GB and I am on a 256kbps ADSL.
On the other hand, if you are expecting that AMD will have advantage in 64-bit mode over Core 2 Duo then you are wrong again. Check this out if you haven’t already:
http://www.xbitlabs.com/articles/cpu/display/core2duo-64bit.html
>A computer that rates a 5 today will rate a 5
>in ten years’ time.
I am surprised that you think so. Sorry Raymond, but I must disagree.
I have a HP laptop which runs Windows XP just fine although a bit sluggish for my taste because I am a developer spoiled with latest and greatest hardware.
Point is that I can’t even install Vista on it let alone think about installing Vista+1. We are talking about the piece of hardware which is barely 3 years old. Can you even imagine how unusable that thing will be in 10 years from now?
>People don’t seem to get confused by the
>fact that CPU speed ratings remain the same
>regardless of which operating system you’re
>using.
It doesn’t stay the same. Newer OS taxes CPU more than the old one. Speed rating is one thing and as such it is useless if you take it out of context. The user experience is what you want to measure and that changes with each new OS and application.
>A graphics card that rates a 3 in Vista will
>also rate a 3 in Vista+1.
How can you claim that?
Consider this example:
I have DX9/SM3.0 capable video card. If it scores 3 in Vista and if Vista+1 uses DX10/SM4.0 then what score will I have with my old card?
I’ll tell you — 0 (as in ZERO), not 3 unless there is a fallback mode to DX9/SM3.0.
But even if you have the fallback, my point is that score of 3 in DX10/SM4.0 mode is not the same as 3 in DX9/SM3.0 mode because it doesn’t translate to the same level of user experience. I sincerely hope you can understand that.
>Having the rating change depending on the
>OS would require software vendors to be
>able to see into the future.
I don’t agree. And here is why:
>This program requires a Vista Medium
>computer or better. If you are running
>Vista+1, call us and we’ll tell you what
>the Vista+1 requirements are
Point here is that program alone will work with Vista Medium capable computer but Vista+1 itself most likely won’t work with such computer.
From the part which says “Vista Medium
>computer or better” it is clear that one who runs Vista+1 will have to have better than Vista Medium to begin with.
By the way, you are completely ignoring the fact that when Windows XP came out many Windows 98 programs stopped working even though they had much more capable hardware work on.
Anyway, trying to make “one rating to rule them all” is ridiculous. Not all applications use the same set of resources. If Average Joe’s computer scores 5 for CPU and 3 for HDD he will get overall of 3. If application vendor asks for overall rating of 3 (aka Vista Medium as you called it) and the application is disk intensive his computer will pass as Medium but it still won’t work as advertised.
To conclude, I stronly suggest using the word rating to judge just the user experience for the current OS and leave the applications alone since they already have their way to judge the required performance level. Not only this new rating system would be redundant but it would be much less accurate because nowadays many applications have benchmark mode where you can test your system using real workload.
Just my 0.02
Igor, as the range is currently "numbers greater than 0" or something like that, ano not "0 to 5", there will never be a need to redefine the number range as including "5 to 10" as the number range already includes 5 to 10.
How about "Requirements: Download the stupid demo and see how well it’ll run on your computer before you shell out money."?
From this article on The Old New Thing Help text is not the place to put logic puzzles. In Windows Vista, when you go to the System control panel, you are shown a number that describes your computer’s rating. But are higher numbers better or worse? If
Thanks to Lilia Efimova , I saw Raymond Chen's news of a book he's publishing in Jan, 2007: 'Old
PingBack from http://blogs.msdn.com/oldnewthing/archive/2007/03/12/1861385.aspx
Dialog boxes are not the place to put logic puzzles.
帮助文档ä¸æ˜¯è®¾ç½®é€»è¾‘迷题的地方。