Log in

View Full Version : stats borked....



Bok
05-03-2005, 03:25 PM
Looks like everyone is in the same boat though...

The format of the text file from the find-a-drug page has lost two columns. I fixed that, but there is a dodgy entry in there too..

7006492,'butch81385',2037,82,893(43786,8747,837,10616,850,1370,295,1115143706

note the '(' in the middle of a field... it's causing the number to be somewhat wild!!

I'll put a temporary fix in for now.

Bok

Bok
05-03-2005, 03:40 PM
ok, temp fix is in....

if ($line =~ /butch81385/) {$line =~ s/\(/,/g;}

:p

Bok

rsbriggs
05-03-2005, 03:42 PM
Must be PHP, doesn't look like awk/sed to me....ZZ

Bok
05-03-2005, 03:46 PM
Perl :thumbs:

rsbriggs
05-03-2005, 04:02 PM
D'oh. I'm getting punchy from these escape key escapades....

Scootenalongnow
05-03-2005, 04:27 PM
FAD Forum (http://www.find-a-drug.org.uk/forums/viewtopic.php?t=5068)

Several different problems going on at the same time I think.

Wizzard~Of~Ozz
05-03-2005, 07:52 PM
Yes, the ( was an interesting thing, I just fixed the file, threw it on my server and told it to download from there.. but I did have to be here for it :S

The issue with the "rank" and "Mystery" column disappearing is one isse, the other was the loss of sort, it looks like we took a few steps back in the evolution of the CSV file..

But yes, we were all in the same boat, same creek, no paddle in sight :) (even the official stats borked on that one)

Only difference, I had to adjust my code when he added those 2 columns, so the native code doesn't need them anyway (I just chuck them out). but that darn ( I guess it could be 1 of 254 possibilities.. too many to filter on a regular basis.. tho I may have a solution, it's doubtful I'll implement it right away since testing could wreck things..

Bok
05-03-2005, 08:06 PM
Hey WoO,

In perl, it just ignores it if you don't get it on a split so the projrank field for me ended up being unset. I just coded a

if (!$projrank) {$projrank = 0;}

and in my config file for fad I set

$doprojrank = 'Y';

and my script goes through and rerank's anyway. I have code already doing that as a number of projects don't provide the rank.. :(

Lengthens the time to run somewhat but nothing major..

Wish they would warn us before changing it though!

:cheers:

Bok

p.s. I don't actually save the file, I just get it directly in perl and parse in memory. I wasa bit concerned that the error might persist on future updates, which is why I coded a temp fix in, you might want to watch the next update coming soon for that.

NiceMedTexMD
05-03-2005, 08:13 PM
What happened to the stats? I'm not even on the stats any more on the homepage for Find-A-Drug. I know it let me download the new version..

I don't have to start over with stats do I? :(

Mike..:Pokes:

Bok
05-03-2005, 08:27 PM
You are still here

http://stats.free-dc.org/new/teamstats.php?proj=fad&team=FreeDC

which means you are still in the stats system at FaD, probably just their html is borked.

So, no, nothing to worry about!

Bok

Wizzard~Of~Ozz
05-03-2005, 08:36 PM
No Nicemed, the problem lies with anyone in the source file that was below butch, his (Official Page) script obviously crashed at the bad input, so since it was unsorted.. many are missing, (Including me) Bok and I have found ways around this, I've got the code written to validate the file, but I will have to measure the performance hit, the stats update is already running 6 minutes through VB, then another 10 minutes through PHP and SQL... (mind you, I shuffle around a lot of data.. in the order of 10Mb/update..)

I haven't look much into perl (looks very close to PHP tho your formatting differs from mine), I run strictly VB for the main pages, tho I may be departing from the main CSV file and just make my own system to build it.. I don't like relying on that many files. so that could be 4000 lines of code in the dumper, I'll see how I feel this weekend.. if it happens then I'll go strictly PHP.. it's a bit slower, but it should be able to flip numbrs around a bit faster using SQL for most of the work.. (VB is Written to use templates and SQL isn't required..)

Bok
05-05-2005, 02:17 PM
Originally posted by Wizzard~Of~Ozz

The issue with the "rank" and "Mystery" column disappearing is one isse, the other was the loss of sort, it looks like we took a few steps back in the evolution of the CSV file..



Hey WoO,

'Mystery' column is actually rank change :)

I don't use it anyway as I doubt you do..

Bok

rsbriggs
05-05-2005, 02:55 PM
When do the BOINC stats change?

BOINC Einstein, last updated at 7:00 yesterday morning, just to mention one....

Bok
05-05-2005, 03:06 PM
Only about once a day, though I check for updates hourly..

Bok

rsbriggs
05-05-2005, 03:23 PM
Hmmm. I see what you mean. E@BOINC has't updated at Synergy for a day-and-a-half....

Ooops - this is the FaD section. Never mind :|ot|: