staff blogs

:: 17-Sep-1999 05:29 (Friday) ::

They say hindsight is 20:20, but they lie. I have no clue what just
happened. The part of stats where the day’s activity is moved from the
“daytable” work tables into the big rc5_64_master table (all 20 million
rows of glory) simply failed to happen.

It silently didn’t append the 16-Sep data to the big table.

Then, of course, it went on to re-rank everything with 15-Sep data.
Needless to say, it didn’t take long for everyone to catch on that
something had gone wrong.

I pulled apache offline and tried the append query manually. worked fine.
I scratched my head, and started all the rankings queries again. They’re
nearing completion right now.

Everything is looking just fine, but there’s still the small matter of not
knowing what went wrong in the first place. In true programmer fashion,
I’m not truly concerned if we can’t repeat the error, so we’ll see what
happens tomorrow. Worst case, we’ll just blame the OS. :)

:: 16-Sep-1999 22:26 (Thursday) ::

See below for more details. I’ve got the “pretty” version of phistory
all polished up and ready for display.

http://stats.distributed.net/rc5-64/phistory.php3?id=1

There’s also a convienent link from everyone’s psummary pages, so the
participants who don’t read the .plans will find it, too.

As daa said, the graphing is mostly functional, although it’s still
slaving from nodezero which is a pessimal solution. I’m also going
to try to come up with a solution that doesn’t involve the mess of
gnuplot that we were using. I should be able to do the graphs in
php which is a much more efficient solution.

then, of course, team histories…

:: 16-Sep-1999 21:21 (Thursday) ::

Well , now that I got Nugget to get phistory_raw.php3 to work.

http://www.distributed.net/statistics/em_plot.cgi?id=1

is back , but will be moving to stats.distributed.net

:: 16-Sep-1999 20:43 (Thursday) ::

Just a quick update, but a pretty momentous one. The first (and most
crucial) step to getting history graphs back online has been passed.

As I’ve explained earlier, the implementation of retiring emails had
significantly increased the complexity (and cost) of running history
queries. This is why we haven’t really rushed to get those queries
re-implemented.

It finally dawned on me today how to accomplish the historical query
at a reasonable speed and still incorporate retired emails into the
equation. This also marks the return of the fabled _raw queries from
the old statsbox.

http://stats.distributed.net/rc5-64/phistory_raw.php3?id=1

As you can see, it’s really not much to look at yet, but your scripts
should love it. Next step is to make a pretty, human-readable version
and then we do graphs. The hard part is behind us at this point.

Time to throw on the headphones, crank up some crunchy guitar and
start coding furiously. :)

:: 15-Sep-1999 22:03 (Wednesday) ::

We just can’t win… I have a confirmed case where an @hotmail.com
participant password was compromised on 9-Sep-1999. I’m aware of th
recent yahoo-reported