RE: [vchkpw] Eureka! Finished POP3-Frequency-Patch (against brute forcing)

2004-02-11 Thread Anders Brander
Hi,

On Thu, 2004-02-12 at 04:40, Jake S wrote:
> >> Also, perhaps instead of "you have to wait xx minutes" maybe you can
> >> just list 0 messages.
> > The idea of listing 0 messages (as new) could lead to some support
> > nightmares. A customer consequently using the wrong password, and there
> > is no sign that anything is wrong - or worse, some third malicious part
> > causing this.
> I'm not seeing your logic if a user has made it to checking their
> inbox then the credentials would have already been checked via vchkpw,
> correct or not and the appropriate errors would be listed.

Oh i see - I thought you meant it should return "0 new messages" for bad
user/password - but you actually meant "0 new messages" as response to
correct user/password, but only after x failed tries?

> Also, with a timeout error code your bound to get support calls asking if
> you can bend the rules for that user because they have a "very" important
> message (usually larger penis ads) verses you simply say no new messages
> and no one knows the difference.

If you just say no new messages, it can go on for month without the user
knowing it. It only takes one malicous attacker x failed authentication
attempt every y minutes to effectively suspend mail delivery.

And instead you will receive support calls/emails that goes like "I NEED
THAT EMAIL NOW!!! MY CLIENT SENT IT LIKE 20 YEARS AGO, AND IT STILL
ISN'T HERE!!! SOMETHING IS WACKED WITH YOU PEOPLE!!" (Yep, smile! :-))

/Anders




RE: [vchkpw] Eureka! Finished POP3-Frequency-Patch (against brute forcing)

2004-02-11 Thread Jake S

Anders Brander said:
> Hi,
>
> On Thu, 2004-02-12 at 02:15, Jake S wrote:
>> Also, perhaps instead of "you have to wait xx minutes" maybe you can
>> just
>> list 0 messages.
>
> The idea of listing 0 messages (as new) could lead to some support
> nightmares. A customer consequently using the wrong password, and there
> is no sign that anything is wrong - or worse, some third malicious part
> causing this.
>
> /Anders
>
>
>
Anders.
I'm not seeing your logic if a user has made it to checking their
inbox then the credentials would have already been checked via vchkpw,
correct or not and the appropriate errors would be listed.

Also, with a timeout error code your bound to get support calls asking if
you can bend the rules for that user because they have a "very" important
message (usually larger penis ads) verses you simply say no new messages
and no one knows the difference.

Just some thoughts

Either way I'd like to see that patch!


Thank you,
Jake S