Change to Scan IDs
 

For people coding external tool sites that wish to parse scans, it is rather useful to store the IDs of scans and scan groups, usually inside a database. Fortunately the biggest number an Integer (as defined by the ANSI SQL standard) can take is 2.1 billion, and so it should be no problem.

Unfortunately it seems that PA uses IDs bigger than this. Why on earth you need more than 2 billion possible IDs I do not know.

In an attempt to provide an open source and easily accessible tools set, this becomes a problem for myself and others. (In order to maintain my database agnostic stance -part of the easily accessible thing- I must accept that there is no such thing as a "BigInteger".)

Please could the code for generating these IDs be modified to resolve this problem? Shouldn't take more than 5 minutes :)

Thanks

Re: Change to Scan IDs
 

As mentioned in #support, we used an unsigned integer.
We use a wide number to make it impossible for an alliance to go through all potential numbers and 'steal' scans.

Re: Change to Scan IDs
 

Thanks for discussion in #support.
In light of Cin's plans for the future I will change my storage strategy.

Re: Change to Scan IDs
 

on a side note, could you get by with just using a string ellonweb? they probably use an integer in order to randomize them. since the number is already generated and won't change you can probably use a string and just concatenate the url to create a link?

Re: Change to Scan IDs
 

Not very elegant though.

Re: Change to Scan IDs
 

Cin's future plan involves switching scan ids to alphanumerics, so yeh I'll use a string. When you think about it, the id doesn't actually represent anything, it is just a key to get at the data so while storing a number as a string is usually a horrible thing, in this case it's not so bad. And I'll be future-compatible.

On a side note Cin said the change won't be made for this round (on scans, anyway), and when it does he'll make sure to give plenty of advance warning for those operating tool sets. :up:

Re: Change to Scan IDs
 

i said it is unlikely that it'll happen for r30, not that it won't, and that i'd announce it to give ppl time to update their tools, nothing about "plenty of advance warning".

Re: Change to Scan IDs
 

ITT, Cin shows how much PA Team cares about its users. Keep up the good work, Cin!

Re: Change to Scan IDs
 

There were people who ran concentration camps who had better people skills than you.

Re: Change to Scan IDs
 

May I ask why you plan on changing to alphanumerics?

Re: Change to Scan IDs
 

I'd guess to make it even harder (re: so very unlikely that alliance techs wont consider wasting time trying) to randomly guess scan-links to 'steal' scans? Liklihood of succeeding in guessing a 10-digit alphanumeric id is same as guessing a ~15-digit numerical one. And shorter ids look nicer. Probably other reasons too :(

I like this move by pa! Don't get why cin is getting mild abuse for it, but I guess that's to be expected from the pa community.

Re: Change to Scan IDs
 

Probability of guessing a 10 character alphanumeric ( case insensitive ) : 1 / 36^10 = 1 / 3656158440062976
Probability of guessing a 10 character alphanumeric ( case sentitive ) : 1 / 62^10 = 1 / 839299365868340224

Probability of guessing a 15 character numeric : 1 / 10^15 = 1/ 1000000000000000

They really arent similar :)

Re: Change to Scan IDs
 

Wow! there's 26 letters in the alphabet... I was using 24, shudda paid more attention to GCSE english innit.

1/ 1000000000000000
1 / 3656158440062976

And they are more than comparable when you're discussing orders of magnitude :( Granted I didn't think of case sensitivity.

Re: Change to Scan IDs
 

If that is really the reason, you're solving a problem that doesn't even exist. No sane human being is going to bother finding scans in a pool of 2^32 IDs, especially when you keep in mind that it serves no practical good whatsoever.

Re: Change to Scan IDs
 

Apparently there have been attempts.

Re: Change to Scan IDs
 

I don't care, it's useless?

Re: Change to Scan IDs
 

I think pa are just trying to get people to stop attempting it (whether they succeed or not, the server is still getting bombarded). So decreasing the chances of guessing correctly by vast orders of magnitude should help deter the idiots.

Re: Change to Scan IDs
 

tbh i would go the other way and generate a fake scan for those bombarding it with urls - just to mess with them.

Re: Change to Scan IDs
 

Afaik (no word from Cin if this is the reason) there is a weakness in the current ID generation, and it is possible to 'steal' some scans in some fashion.

Re: Change to Scan IDs
 

it boggles the mind as to how on earth something which should be as simple as a random number can be implemented wrong

Re: Change to Scan IDs
 

My guess is that they reseed it on every call with the current time.

Re: Change to Scan IDs
 

no we don't, it's a totally randomly generated numerical value, or was anyway.

ID change has been made, it applies both to scan and stored news IDs.
Both are a 15 char long alnum, so the urls and bbcode tags look like this:
show_news.pl?id=4s7hs81rladx2au
showscan.pl?scan_id=3vfbj73gleoihab
[scan]3vfbj73gleoihab[/scan]

Re: Change to Scan IDs
 

those samples dont look case sensitive to me, does the implementation include capitals as well?

Re: Change to Scan IDs
 

You might want to read Knuth and then reconsider whether a "random" number is simple or not. Granted, in this case there are quite a few less constraints on the "random" number's quality, but ultimately it is not easy to get good "random" results.

I recommend you read up on how "random" numbers are computed.

Re: Change to Scan IDs
 

Im assuming they are using perls rand() funct , which seeds on first use if not done already. Its hard to mess up if you are using a single function call without deliberately sabotaging your own efforts by reseeding
http://perl.active-venture.com/pod/func/srand.html
On the servers, /dev/urandom should be available - plus the process id and memory allocation states should not be externally available to compromise the entropy pool

Re: Change to Scan IDs
 

Thanks for quoting the perl documentation, now go back and read Knuth to understand what I tried to tell you. It's in Volume Two of TAOCP.

Re: Change to Scan IDs
 

Thing is though, even if it's not perfectly random, it doesn't really matter, because the vulnerable information is of little to no value. So instead of wasting time on coding shit like this, perhaps it would be a good idea to,oh, I dunno, improve the game?

Re: Change to Scan IDs
 

I hope this was done with good reason. It sucks.

Re: Change to Scan IDs
 

Agreed tbpfh.
"ZOMG I stole a scan done 4 ticks ago on a planet with 2 distorters, I'm so l33t"

Re: Change to Scan IDs
 

it sucks? it's about 1 minute of recoding to fix a parser....
I don't see the problem.