100K

[MarkV]

It's the same with IPs, today I use one, tomorrow another. Nothing won, nothing lost.

But think about routers. 10 PCs with HFS behind a router would all have the same IP, but using the ID you could distinguish between them. Same with all NATed networks.

[SamePaul]

Rejetto, it's much simpler.
Once HFS started it checks for existence of ID in local configuration file. If it's found - use it. Otherwise generate new ID based on IP and local time with precision upto microseconds (windows clock allows it as I know). It would be 64-bit integer (32 for IP and 32 for time), but with very high probability such ID will be unique for each user and will keep between sessions and version updates. This ID can be sent in HTTP headers as part of update process.

Btw, I noticed in the firewall logs that hfs periodically tries to connect to 3 different addresses. Is it really necessary?

[rejetto]

generating a unique ID was already suggested.
it would work.
the problem is: would it make people angry?

that connection is "find external address" and it is done every minute.
that's because you use a dns service, and HFS needs to know when your router IP change, to update the dns.
i will lower the frequency to 10 minutes.

[MarkV]

Well, if only counting users per month, it would be easy:

-Check in .ini file: Have I been counted this month?
-If no, generate ID, send ID, delete ID
-Write in .ini: Have bin counted:Yes, if successful

Every HFS would be counted only once per month.

Permanent ID is a no-no.

[SamePaul]

generating a unique ID was already suggested.
it would work.
the problem is: would it make people angry?

Angry? If you publish in "updates" that Unique ID was implemented. People will know why and what for it was done. Also on ID you could do MD5 on final ID so it will never give users IP (however I see it pointless since user IP disclosed already anyway when he connects to update service). So anonymity (if someone concerns) would stay untouched.
And those who are really paranoid can turn off automatic update checking (possible already) and this way no information would disclose at all.

[SamePaul]

Well, if only counting users per month, it would be easy:

-Check in .ini file: Have I been counted this month?
-If no, generate ID, send ID, delete ID
-Write in .ini: Have bin counted:Yes, if successful

Every HFS would be counted only once per month.

Permanent ID is a no-no.

Generating+deleting ID has no point since it is just like without ID at all. Even worse - same IP will give different ID every update and thus counted many times.

[MarkV]

Well, if only counting users per month, it would be easy:

-Check in .ini file: Have I been counted this month?
-If no, generate ID, send ID, delete ID
-Write in .ini: Have bin counted:Yes, if successful

Every HFS would be counted only once per month.

Permanent ID is a no-no.

Generating+deleting ID has no point since it is just like without ID at all. Even worse - same IP will give different ID every update and thus counted many times.

Please read it again. Every HFS would only get counted once a month. There is no need for the ID anymore. It does not matter if the ID is different the next month as it would be a new count.

Got it? Each HFS would get counted once a month and sleep the rest (in regard of counting). That post was made under the assumption of only counting users this month.

[SamePaul]

Please read it again. Every HFS would only get counted once a month. There is no need for the ID anymore. It does not matter if the ID is different the next month as it would be a new count.

Got it? Each HFS would get counted once a month and sleep the rest (in regard of counting). That post was made under the assumption of only counting users this month.

no, i don't. How it comes that user will count once per month. You propose that update checking will be once per month? Explain your algorithm it's not clear.

[TCube]

Each HFS would get counted once a month and sleep the rest (in regard of counting). That post was made under the assumption of only counting users this month.

Reading this of a while 
I'm not picking on U MarkV  ... but this silly thing about "best way of phoning back home" is still .... "phoning back home".
I'll developp later during the day in my best possible english   
TCube

generating a unique ID was already suggested. it would work. the problem is: would it make people angry?

You bet 

[MarkV]

I think I'll stop it now. Just one final word: Please assume you would restart the count from zilch every month, thats what I meant.

[rejetto]

Having an ID that is different month-by-month is a technically ok, i guess.

It's only a matter of adding a thing that users may think bad. It's a risk.

At this moment HFS doesn't send anything except what is needed for a normal http connection, and we have still a decent counting.

[Tuskenraider]

you know... tracked or not... i say have hfs generate a random value. ip.ip.ip.value. if hfs has taught me anything its that everything we do, look at or download on the internet is tracked. so ill put money on the fact that the people that would feel tracked..... already are..
and one more idea have a autocheck marked function for hfs reporting, and allow people to opt out of being counted (tracked).. ie no ip ip ip.value being sent..

idk just a thought...

tusken

[jack_2000]

Simple, Send the mac address
Afterall mac's are not a secret.

[rejetto]

I never said an ID would be a evil thing.
I just said people would perceive it as evil.
There are tons of paranoids out there who would stop using HFS and advice people against it.

[Fysack]

WOW! 100k that a lot man. Glad i didnt miss this You are the king rejetto. You, and your insane posters.

You know the dude that invented chess, he give it to the emperor as a present (have i told this one before? ) The emperor said; I love this game, ask me anything you want, and i will make your wish come true. "Thats easy" said the  chessinventordude, just give me 2 rice-corns for the first square, then you dobble it for the next, and for the next you dobble it again, and so on. The emperor sent for his mathmaticians and thinking, this was a easy one. ..just to find out that this much rice did not exsist in the whole empire.

My point is; 100k is a hell of a lot more than 50k. A lot more than twice as much  Capice