Functional requests

[Yobgod]

Hey all... I have some free-ish time and decided to start playing around here.

I successfully confirmed nine answers to early problems, but now it seems that any submission I try on any number of other problems spits back to me with:
The confirmation code you entered was not valid. Please try again.

I am definitely entering the correct confirmation code, and can provide any number of screen captures to verify this.
According to the old news items, I also shouldn't be anywhere close to my "anti-cheat" limits.

Is anyone else having similar problems? Does anyone have any suggestions for how I could get this debugged?
It may be an intermittent problem, as I was unable to submit one answer, but after trying 8-9 times it finally went through (and was correct! ). So far no amount of resubmitting has helped with the problems I'm currently trying to submit.

[daniel.is.fischer]

I think the CAPTCHA expires after a while, so if you let some time pass between loading the page and submitting, that may be it.
Does refreshing the page (should bring up a new CAPTCHA) immediately before entering the answer and CAPTCHA help?

[Yobgod]

No help there.

I submit within a second or two of the CAPTCHA image loading (which itself takes 3-4 seconds), so unless the expiry occurs in a relatively small number of seconds that shouldn't be the problem.

With my answer in my paste buffer, I just tried submitting problem 9 twenty times in quick succession... all failed with the same "confirmation code not valid" message. *sigh*

[daniel.is.fischer]

Pity. Would have been nice if it had been something simple.
I've no idea what it might be, so I can just throw random ideas at you.
Maybe it would work with a different browser (not uncommon for things to work in one browser but not the other).
You're not having spaces in the confirmation code, have you?

[Yobgod]

It does appear to be OS/Browser related.

I was having the troubles on CentOS5.3/Firefox3.0.12.
If I write down my solutions and go find a Windows XP box with IE7, I can enter the answers successfully.

[daniel.is.fischer]

It does appear to be OS/Browser related.

I was having the troubles on CentOS5.3/Firefox3.0.12.
If I write down my solutions and go find a Windows XP box with IE7, I can enter the answers successfully.

Perhaps a less drastic measure would suffice?
New version of FF, SeaMonkey, Opera?

[Yobgod]

Problems now seem to be working more often than not, so I'm generally happy (if still puzzled).

The weird thing is that it really seems to be a per-problem issue... if a particular problem submits, it submits reliably, and if it won't, it reliably won't.

Since I can find workarounds, it's probably not worth pursuing, especially if no one else is effected.

[gfunchess]

It would be nice to see stats like standard deviation based on the number of problems solved.

[titanix88]

How about some criteria of some(if not every) problem being randomly generated? Like for the problem 6, instead of 1 to 100, some users will be shown (2 to 56), (1 to 51) etc. This will reduce cheating greatly. At least everyone will be forced to run a program (who knows, maybe then someone will create a web service to solve this problems automatically :-/). Such dynamic problem-solution systems exist e.g. www.hackthissite.org which is a computer security related educational site. Especially the "application programming" section there generates problems on the fly and answers are valid only for a certain amount of time.

By the way, thanks for creating this site.

[PhilTaylor]

It'd be nice to have some sort of API that I could use to link my ProjectEuler stats up to my own website - even some sort of RSS for the list of solved problems by a specific user would be nice

[Georg]

It'd be nice to have some sort of API that I could use to link my ProjectEuler stats up to my own website - even some sort of RSS for the list of solved problems by a specific user would be nice

Perhaps you could use this.

[mrtollefson]

I'd like a cookie or server-side file that saves the preferred sort of the problem discussions. Similar to the feature on the main PE site.

Thanks.
...mrt

[elr]

what i would like to see is turning user names at problems forums into links (so you can click on the user and see
his profile) ,the same for flags for flags

[sfabriz]

Can I suggest one thing here?
The "view all of the problems on one page" has now become incredibly huge.
For those of you like me that love to have a pdf of all the problems, the document is currently 121 pages for 2.5 Mb.
Could it be possible to insert one step between the "view all the problems" link and the actual page?
This way you could give users the chance to select a range (e.g. "all", "1-50", "51,100", etc.).
In this way every user who likes to have pdf of all the problems could have something like:
pe_50.pdf
pe_51_100.pdf
...
pe_251_265.pdf (as 265 is now the last one).

I don't know how often that page is accessed, but giving ranges would have 2 benefits: people like me would print pdf quickly (because I'd print just the last one) and would reduce traffic on the site, since you'd spit only last problems page in most cases.

Cheers!

[hk]

This page isn't often visited by Colin, who does the maintenance of the site.
It's a nice suggestion, though.

In the meantime I checked my pdf converter.
When I give it the filename of an existing pdf file it askes me what to do.
One of the options is: append to existing file.
So one thing you could do is append new problems to an existing pdf simply from the problems page of that new problem.

Another option I tried is:
when I print the "all details...." page I can give my browser the page numbers it should print.

Nevertheless your suggestion would obviously be the best solution.

[sfabriz]

Well, yes, generally speaking, the webmaster should never force users to use tricks to get things done.
Anyway, there's no hurry of course, but I think it would be something nice to add.

Cheers
sfabriz

[hk]

Just to work out the idea further:
The "view all of the problems on one page" is part of the static data. Those are cached on the server.
User queries would require intermediate database access. Static data were introduced to minimise the load on the database server which is more of a bottleneck than bandwith is.

The best I can think up currently is to split that page in parts.
Say: 1-50, 51-100 etc or something like that, but no user defined #n-#n+x
Would that be enough?

[sfabriz]

Yes, that'd be good.
Same function (static data) but split into different pages (All / 1-50 / 51-100 / ... / etc.).
This way saving the last segment as a pdf will be faster and also you'll spare some traffic.
Cheers,
sfabriz

[AndreasEisele]

This is not really a functional request, rather a question about a function that already exists, but I didn't find a better place to post.

On the statistics page, the programming languages are somehow ranked, but it is not clear to me according to what criterion. Perhaps some combination of popularity and success rate? I'd love to know the formula that is used for this, so that I can see what to do to push Python on rank 2

Thanks a lot, Andreas

[stijn263]

I think it's: (% solved) * log( number of users )