|
Ok I get it, the non-standard loop threw me off. Your index is also "one lower" so it's also slist[scount] and rnd.Next(0, scount + 1). It just depends on whether scount starts at the length of the list or at the highest index. The correct thing to combine with `rnd.Next(0, scount) is slist[scount - 1] but then of course the index needs to be one higher.
|
|
|
|
|
If you're seeding off of a time, you've got a much more serious problem than just a low number of bits of entropy. You're using a predictable seed which means an attacker can guess it relatively easily and pwn you much faster than brute force.
At an absolute minimum, use a Cryptographic RNG to seed Random. You could use RandomNumberGenerator directly everywhere; and for crypto need to if you want to be secure instead of a subject of lulz on hackernews a few years from now when you're epicly pwnd. But since it only outputs byte streams using it to create arbitrary integer ranges or floats is a PITA.
MSDN is lacking in details (shocking), but presumably to be cryptographically secure it seeds itself using the hardware RNG build into CPUs for the last 10 or 15 years (driven by thermal noise but too slow for use in anything that needs a lot of values).
Did you ever see history portrayed as an old man with a wise brow and pulseless heart, weighing all things in the balance of reason?
Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful?
--Zachris Topelius
Training a telescope on one’s own belly button will only reveal lint. You like that? You go right on staring at it. I prefer looking at galaxies.
-- Sarah Hoyt
|
|
|
|
|
Dan Neely wrote: If you're seeding off of a time, you've got a much more serious problem than just a low number of bits of entropy. You're using a predictable seed which means an attacker can guess it relatively easily and pwn you much faster than brute force.
I'd be very interested to hear more on this.
thanks, Bill
«While I complain of being able to see only a shadow of the past, I may be insensitive to reality as it is now, since I'm not at a stage of development where I'm capable of seeing it. A few hundred years later another traveler despairing as myself, may mourn the disappearance of what I may have seen, but failed to see.» Claude Levi-Strauss (Tristes Tropiques, 1955)
|
|
|
|
|
If your seed is time based, an attacker only has to test the seeds from potential periods when it was restarted instead of all possible values. If you use a single Random app wise and initialized on startup or first use then by watching for an indication of a restart they can often narrow it down to a few minutes or an hour that was used for the seed. And even if they can'd do that if it's a Windows server it's probably been rebooted in the last month for patch Tuesday. Even if you're not rebooting the OS version can tell a lot. eg Windows Server 2016 only came out in September, so you've got a year of timestamps to work with. Finding a bug that lets you crash the server lets you narrow the start time, and thus the seeds you need to test to a few minutes window.
For the specific example in the thread new Random(DateTime.Now.Millisecond); there are only a 1000 possible seeds 0-999; testing all of them until you find the one that generates a random sequence which matches your output is trivial since you've only got 10 bits of entropy.
Did you ever see history portrayed as an old man with a wise brow and pulseless heart, weighing all things in the balance of reason?
Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful?
--Zachris Topelius
Training a telescope on one’s own belly button will only reveal lint. You like that? You go right on staring at it. I prefer looking at galaxies.
-- Sarah Hoyt
|
|
|
|
|
Thanks, Dan, for the fascinating reply. Usually I use 'Ticks:
Random rnd = new Random((Int32)DateTime.Now.Ticks);
not 'Milliseconds, for a seed ... an old habit: I can't even remember why I started doing ... I believe that's probably a waste of time since, I read, the .NET implementation (now ?) uses 'Ticks by default.
cheers, Bill
«While I complain of being able to see only a shadow of the past, I may be insensitive to reality as it is now, since I'm not at a stage of development where I'm capable of seeing it. A few hundred years later another traveler despairing as myself, may mourn the disappearance of what I may have seen, but failed to see.» Claude Levi-Strauss (Tristes Tropiques, 1955)
modified 25-Oct-17 12:49pm.
|
|
|
|
|
That'd be marginally less bad in that Ticks is higher in resolution, but is still approximately predictable based on knowing when the app started.
Ultimately it's MS's fault though for:
0) Not having a crypto secure source for a seed in .net 1.0
1) Not having the courage to break the legacy default behavior in 2.0 by having their shiny new crypto RNG at least seed Random()
2) Not making the Crypto RNG classes feature rich enough to be dropin replacements for the legacy class.
There are cases were all you need is good enough randomness and an RNG that runs an order of magnitude faster than a cryptographically secure one is preferable; but outside of a hotloop that should never be the case. Because developers are lazy and stupid and lazy and copypasta addicts and lazy the default Random() should be a cryptographically secure RNG seeded with a hardware RNG. For cases where the exception is needed both Random and FasterInsecureRandom can both implement IRandom for ease of swapping the latter in.
Did you ever see history portrayed as an old man with a wise brow and pulseless heart, weighing all things in the balance of reason?
Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful?
--Zachris Topelius
Training a telescope on one’s own belly button will only reveal lint. You like that? You go right on staring at it. I prefer looking at galaxies.
-- Sarah Hoyt
|
|
|
|
|
If the shuffle has to be cryptographically strong then System.Random is not an option at all anyway..
|
|
|
|
|
I have to admit that I'm guilty of this... I need to solve a problem, so I Google, find a working solution and paste it into my code. It works, but you don't necessarily understand the solution. On the one hand having answers so easily available speeds up development, but on the other it makes for developers that don't really understand what they are doing?
|
|
|
|
|
All you need to do is spend some time looking at "Q&A" to answer your question.
| Ravings en masse^ |
|---|
| "The difference between genius and stupidity is that genius has its limits." - Albert Einstein | | "If you are searching for perfection in others, then you seek disappointment. If you are seek perfection in yourself, then you will find failure." - Balboos HaGadol Mar 2010 |
|
|
|
|
|
I have to admit that I'm guilty of this... exactly like you!
... such stuff as dreams are made on
|
|
|
|
|
Jacquers wrote: It works, but you don't necessarily understand the solution.
There's two categories. First, are code snippets for doing something I'm too lazy to figure out on my own and/or don't want to read through reams of MSDN documentation to figure it out. Those code snippets tend to be no more than 50 lines long (usually 1-5 lines) and I always look at the code to understand it, particularly as there are minor tweaks that often need to be made, anything from styling to turning it into a more general purpose function that I can add to my toolchest.
The second category is "this is a cool package." Examples include FluentMigrator, NewtonsoftJson, FluentCommandLineProcessor. This is the "copy and paste" an entire project into my code base. No, I don't use Nuget unless absolutely necessary. Any open source package I clone and make sure that I can build it from the source. If it has too many warnings, doesn't compile with C# 7 / .NET 4.7, builds only with x86 rather than "Any CPU" or has other issues, it's rejected. Once it passes that basic test, I don't look very carefully at the source except perhaps out of curiosity because it's too much effort to get into the thinking of the developer, however I do peruse the file structure and class structure to get a comfortable "ok, this looks well thought out" feeling.
|
|
|
|
|
I have to admit that I'm guilty of this... I need to solve a problem, so I Google, find a working solution and paste it into my code. It works, but you don't necessarily understand the solution. On the one hand having answers so easily available speeds up development, but on the other it makes for developers that don't really understand what they are doing?
Did you ever see history portrayed as an old man with a wise brow and pulseless heart, weighing all things in the balance of reason?
Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful?
--Zachris Topelius
Training a telescope on one’s own belly button will only reveal lint. You like that? You go right on staring at it. I prefer looking at galaxies.
-- Sarah Hoyt
|
|
|
|
|
Well done sir  
|
|
|
|
|
I think it's fine in certain circumstances. If I need to do something that's way outside my usual scope and not something I'm likely to ever do again, I'm perfectly happy to "borrow" a few lines of code but I still like to refactor it a little and tidy it up along the way. That way I might not be getting a particularly deep understanding of the code but at least I'm figuring out what's doing what.
If it's something that I do need to know, my preferred method is to just read a few code samples and then write it for myself.
I do despair when I see people paste in what is obviously meant to be illustrative code ("myConnectionString", "yourObject", "//Do whatever you need to do here ..." etc.) and leave it in its original form.
98.4% of statistics are made up on the spot.
|
|
|
|
|
Me copy and paste code? I'm so hardcore I disable Intellisense.
|
|
|
|
|
You use an IDE? Sh*t, I used to write code in EDLIN. 
|
|
|
|
|
EDLIN? Back in the day they only used zeroes, ones were too expensive.
|
|
|
|
|
The problem is not to Copy-Paste or not (I do it too)... The problem is accepting code into your application, you do not understand fully...
To look up a solution to a new problem is probably the best way to learn, but to not research it to the very end is a crime!
Skipper: We'll fix it.
Alex: Fix it? How you gonna fix this?
Skipper: Grit, spit and a whole lotta duct tape.
|
|
|
|
|
I will use a "developed control" without looking too hard (or even at all  ) at the source code, provided it's from a source that I trust.
But I won't copy'n'paste code - even small snippets - without understanding how it does what it does. If I don't, then I'm in the position of "then a miracle occurs" and I can't fix it in response to changes, or do it again next time something similar occurs.
I do not understand the "grab random code from the internet and hope it works" mentality - it just seems counterproductive in the long term to me.
I don't think it does "speed up development" - it gets you out of a hole right now, but drops you in a bigger one later on!
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
OriginalGriff wrote: I do not understand the "grab random code from the internet and hope it works" mentality - it just seems counterproductive in the long term to me.
I think that may stem from the mistaken belief that there is only ever one way to do something. A sure sign of someone who doesn't have much, if any, experience understanding algorithms and/or writing code.
|
|
|
|
|
I know what you mean, and that just makes it even weirder - since it seems to be a "yoof thing" and they grew up in a much more "no-wrong answers" education system than we did. Or at least I did - there was one right answer and you were supposed to know it rather than work it out for yourself in pretty much all subjects.
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
The absolute plethora of "What is the exact/right way to do this thing" questions on here, or everywhere else for that matter is perfect evidence of this.
I call it the "One Solution to Rule them All" approach, most others call it only knowing how to use a hammer.
Even more so, I laugh at the abundance of "Exam Cheat Sheets" you can easily find online now too, just go download one, memorize the thing, then pass with flying colors the next day, none of this "OH SH*T My final exam is due tomorrow, last minute studying and panic anymore ", I passed my UNI course back in the day... just. or so I'm told.
STILL CRAZY - The BEST and only way to be!
|
|
|
|
|
I had the benefit of a Navy Nuke education in the 70s when Rickover was king. If you could not, on the spot, explain the how and why of every aspect of any process in the nuclear power plant, you were not fit to serve as a nuke.
We were taught to figure out the "how and why" of everything we did, and I carried that educational philosophy over into software engineering.
|
|
|
|
|
I do the same. If I understand the code, I type in a subset of the online version without all the dead code padding that invariably will be there.
As an example, in Winforms when opening a connection, most code wraps the implicit SqlDataAdapter.Fill with explicit conn.Open and conn.Closed. I am fine with just the implicit.
Of course, I also document my code, transfer Class information to Word documents for easy re-accessing, and test the bejabbers out of it.
modified 25-Oct-17 9:25am.
|
|
|
|
|
As others say I'll also copy/paste code, clean it up, but I always take the time to properly understand [and improve to better fit the situation]. I'll never accept black boxes inside my source.
If it's a somewhat significant "borrow" I'll leave a comment/partial url in my source referring back to where I found it and the authors name - even if I subsequently / substantially rewrite it.
Why? starting from real/complete and working example: can try it, test it, and see what breaks it.
once it's initially working then I'll refer to the dry msdn docs to fully understand the methods / return values / got-yas etc.
After that very often I'll do a rewrite (and re-test) now that I have the "inner-mechanicals" in my head - to ensure I really do understand what's happening (and just in case it later breaks).
Installing Signature...
Do not switch off your computer.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
