Hi i just activated my old account. In my account settings my original sign up date and time was taken over wrong:
The seti2-page shows

SETI@home member since 12 May 1999 20:05:07 UTC

But the correct date was

Registered on: Sat May 15 01:05:07 1999 UTC

So why is there this difference?
And is there a possibility to correct this?



\/
Urs

I have the same problem here and my totla work units results are wrong too. :(

I have the same problem.
I registered seti classic on Dec 2 15:32:39 2001 UTC and after I activated my old account the Seti/BOINC stats say I registered on 30 Nov 2001 11:32:39 UTC.
It seems 53 hours were dropped from our reg dates.

Regarding the work unit totals, they took a snapshot of the classic database around May 15.
So your total in the new stats is from that date.
Murphy say's
complex systems cause complex error's
and
simple systems cause complex error's

> It seems 53 hours were dropped from our reg dates.
>

You mean 52 hours in your case, and 53 hours in my case(daylight-saving-time).
"Looks it is a far more complex bug."

The results snapshot i have read somewhere else was taken on May 14th, but the number of results it shows in my case is from May 13th. Maybe i uploaded after the snapshot that day.

>Murphy say's
>complex systems cause complex error's
>and
>simple systems cause complex error's

somebody said: humans learn with their errors


\\/
Urs

> I have the same problem here and my totla work units results are wrong too. :(
>
>

I'm having the same problem now and it's two years later.......

It, more than likely, has to to with a bad date converions between classic and boinc...

Two days off, and two leap days between 1999 and 2004..

Most *nix boxes use the "seconds since "Jan 1,1970,0000 UTC" as the time keeper, and somewhere along the line, the "conversion routine" forgot that "Feb 29" does occur "almost" every 4 years.

Yup, it can be on the order of 50-52 hours, depending on DST too!

Not a big bug, but does show "the science" kind of took a back burner to releasing code!

Same problem :-/
My Sign up date is wrong and total work units results too.

NeoTime

Hi! My registration date/class are important to me too. This is a data base management issue I think rather than a date conversion issue. The original date and time were correct. Why should they be have transposed or converted at all? They just needed to be retained accurately. I'm OK because I have my own accurate records and I have snapshots of my class records but I would like to have it right on my profile... Mine was "Sat Jul 17 17:31:33 1999 UTC" My first WU took about a week to finish. :-)

This is a database management problem. Why should there be any date/time conversion issue? The original registrations were correct. They only needed to be retained. :-)

> This is a database management problem. Why should there be any date/time
> conversion issue? The original registrations were correct. They only needed
> to be retained. :-)
>
>

The data in the Clasic DB is stored internally in "some format", and it could very well be the "seconds since 1/1/70 format used by Unix (a 32 bit unsigned int) (that DB is informix, IIRC, and maybe pre Y2k!). When exported, it might be a "string" (eg "Feb 27, 2000 17:34")

When imported into the new DB (mySql for BOINC), the string may or may not be converted back into a "seconds since 1/1/70" format" (new version of Unix epoch is now > 32 bits, as the 32 bit version will "wrap" in 10-20 years. (The son of the Y2K bug that we all worried about 4 years back, but has yet to make the press!)

So, it does seem someone forgot about "Feb 29th" along they way.. MySql is MyGuess, but it could be that classic has an old informix

Case in point from the Classic Forums:

"... and registered on 24/08/2000 but the web interface reports that with BOINC i'm was registered on 23 Aug 2000 "

Note in THAT case there was only one "leap day", as the user didn't join until after leap day in 2000, so is therfore only off by one day!

> > This is a database management problem. Why should there be any
> date/time
> > conversion issue? The original registrations were correct. They only
> needed
> > to be retained. :-)
> >
> >
>
> The data in the Clasic DB is stored internally in "some format", and it could
> very well be the "seconds since 1/1/70 format used by Unix (a 32 bit unsigned
> int) (that DB is informix, IIRC, and maybe pre Y2k!). When exported, it might
> be a "string" (eg "Feb 27, 2000 17:34")
>
> When imported into the new DB (mySql for BOINC), the string may or may not be
> converted back into a "seconds since 1/1/70" format" (new version of Unix
> epoch is now > 32 bits, as the 32 bit version will "wrap" in 10-20 years.
> (The son of the Y2K bug that we all worried about 4 years back, but has yet to
> make the press!)
>
> So, it does seem someone forgot about "Feb 29th" along they way.. MySql is
> MyGuess, but it could be that classic has an old informix
>
> Case in point from the Classic Forums:
>
> "... and registered on 24/08/2000 but the web interface reports that with
> BOINC i'm was registered on 23 Aug 2000 "
>
> Note in THAT case there was only one "leap day", as the user didn't join until
> after leap day in 2000, so is therfore only off by one day!
>
>

A valid point, and I'm no expert in the UNIX date formats, or storage of them. However, I can only point out that my registration date was Sun Dec 1 02:57:36 2002 UTC, and I am now shown as 28 Nov 2002 22:57:36 UTC. Thus a total of 52 hours, despite there being only one leap day between the two dates.

I have to say that there does seem to be a general opinion that there were 52 hours "dropped", but your reference to a member who only lost one day, I don't understand.

Until we have a techy from SETI find out what the issue is, and tell us all, I think we can speculate till we're blue in the face - but I doubt we'll get to the bottom of it!

Of course, it's possible that any or all of the explanations provided by good people like yourself are correct, but that for some reason different errors were encountered on different batches of member transfer. I am assuming that since there are over 5 million of us, that the data transition would have been split up into smaller batches for "conversion" to the new database.

However, I think the basic concept of the leap day is most likely to be at least part of the casue, and I am also assuming that there was no "conversion", but simply a transfer of the raw data which is now being mis-represented on the new system. This could easily be at least related to the technical bit which you have explained above.

Finally, with regard to the "son of the Y2K bug"... I am intrigued! I had forgotten about that, and I can't remember now when it's due to hit - I have a feeling this may be 2012... anyone know the theretical problem date?!??

Rookie.