"ReleaseDate" - > "Released" in CLTBoss output is done. Was passing thru that code, so I went ahead and made that change.

Sounds like you're close to the finish line (knock on wood). It's been a long and ardous journey. I'm glad that you stuck it out. 

Really looking forward to running some test in order to tweak CLTinfo.

Quoting mediadogg:

Quote:
Do you want a link to the Christopher Lee Invelos XML?

Thanks, but that will not be neccessary. I'm now more interested in variety that volume.
I took my collection export file and added two variant nodes to it. 8800 profiles loaded and analyzed in 13 seconds. Given that even the worst cases, like Christopher Lee, doesn't come to half that, I think I don't need to spend my time trying to speed that up.

I'm interested in files where there is a mix of F/M/L and Credited As hits, to make sure I count the credits correctly. But that can wait until you have a version that you feel you can share, so I can experiment myself. My bit is a walk in the park, compared to the complexity of CLTBoss, so I don't want to burden you with stuff I can do myself.

Danny Devito kicking my butt big time. Perfect scrape of 2560 UPCs from CLT. However my credits scan picks up only 2020 profiles!!!!!!!!!!! Why is it that I get perfect results for actors with hundreds of profile credits, and get errors on guys with thousands????       

Makes it harder to debug. Oh well, enough whining ... back to work.

Oh wait, what if I scan in batches of 100. Could it be a memory issue? Hmmm ...

The good news:
- beginning to see stability. Get same results for same actors, every time

The bad news:
- there is a subtle problem that causes me to consistently get the wrong answer for certain actors

I have a set of actors that I use for testing that span from butt easy as a quick check, to difficult due to size, complexity of credits, corrupt profiles, etc. I constantly re-run the same ones to make sure I haven't broken something or to see if I have fixed the known errors.

For example, these all run totally 100% correct each time:

- queen elizabeth
- donald trump
- will wallace
- and many others

But for reasons I am still figuring out, I get consistently the same shortage of profile hits for:

- danny devito
- tom cruise

Cross fingers, I have been getting 100% scrapes for some time now. But there is something with the XML Xpath scan. I am just talking out loud to blow off some frustration. I am not actually asking for any specific help at the moment. I suspect that I have error trapping code that is hiding an error of some sort that causes the skipping of certain profiles. Could be an Xpath error, unexpected characters in the datastream, out of memory error, something like that ... (and oh yes, I have an error log.)

It wouldn't hurt to have CLTPlus output for Danny Devito, if available. It would be gravy - I probably have what to need to solve it without that additional data,

Thanks to my fellow soldiers. Hopefully I will not have to request this again. With the ones I have now, and the debugging tools I have been gradually accumulating, I should be able to figure out what is going on.

To give you a flavor, yesterday, I was suddenly getting a weird error when I was loading and saving past results in CSV format. Turns out the title of some movie in some language I don't know, had a ";" in the title as part of the way that character set is represented in non-unicode. But that character is deemed to be a  delimiter, so I had to modify my error trapping to let that one pass, in this circumstance.

Edit: FYI. "Páginas de Liberdade (Freedom Writers)"  (holy smokes did you see that? got translated to A tilde i - but when loading the raw data, I assure you there was a ; in there - WTF?" (CSV processor shares my bad eyes. The character is not ";" but also not "i" on the same code page.)

Quoting AiAustria:

Quote:
you have to type the www. in front of the domain: http://www.dvdprofiler.com

Ah, maybe that's where I went wrong. Thanks!