14'43"

Requesting CLTPlus of ziyi zhang (365).

I am always getting a couple extra. I hope it is something I can fix (which might also fix other things), and not more database weirdness.

Quoting AiAustria:

Quote:
Done.

This should be an easy fix, as soon as you help me understand how.

Regarding "CreditAs", there are 4 cases:

(1) CreditedAs field is blank - no issues. Can't use what is not there.
(2) CreditedAs field not blank, but does not match variant being searched. No issues. No match on it.
(3) CreditedAs field not blank, does match the searched variant and FML does not match. No issues, a match.

(4) CreditedAs field not blank, does NOT match searched variant, but FML DOES match
  - CLT does not count this
  - CLTBoss (as of now) counts this

Should I follow the CLT method?
Would it be helpful to have an option to count an FML match, even though non-blank CreditedAs does not match?

Quoting AiAustria:

Quote:
Quoting mediadogg:

Quote:
Should I follow the CLT method?

Yes, the CLT is referenced in the rules as source for common name decissions.

Quote:
Would it be helpful to have an option to count an FML match, even though non-blank CreditedAs does not match?

I personally don't see a valuable information in "Which profiles exist with this F/M/L in the online data base?" (sound like a IMDb query for me) - but maybe there is one, which I can't imagine right now...

Already coded and tested in anticipation of that answer. 

Quoting AiAustria:

Quote:
Quoting mediadogg:

Quote:
Should I follow the CLT method?

Yes, the CLT is referenced in the rules as source for common name decissions.

Quote:
Would it be helpful to have an option to count an FML match, even though non-blank CreditedAs does not match?

I personally don't see a valuable information in "Which profiles exist with this F/M/L in the online data base?" (sound like a IMDb query for me) - but maybe there is one, which I can't imagine right now...

Well, it might be interesting to be able to compare the CLT info with the F/M/L info for statistical purposes. If, for example, the two are very close, then doing research into actual credits becomes more important than if there is a clear lead for the CLT number.

Wow, Christopher Lee 3516 / 3516. Gonna quit work and have ice cream and a movie now.

Edit: Tom Hanks 3887/3887

That is a weird one! if you check http://www.invelos.com/Forums.aspx?task=contributionnotes&type=DVD&ProfileUPC=5051891151710.13, you'll see 2 earlier contributions.

However, when you set up a profile with that UPC+locality Italy, and contribute, you'll get to a blank contribution page, as if you're the first one.

I'll recontribute with the info from the alternate profile, and see if the original is 'reset'.

I am tempted to release code soon. It is unrealistic for me to even try to write some kind of user manual. First, I don't have the time. Second, the people that are likely to use CLTBoss are already smarter than me, and will figure things out, or complain enough to get my attention.

So, I plan to make a few video tutorials instead.

Given the technology and my skill set, I will offer three methods to get the fastest most accurate results, and make a video example of each:

(1) Manual web browser "mouse-click" scrape of CLT: followed by automatic scan of online profiles. In my testing, this is by far the fastest and most accurate method, with the downside of requiring one mouse click per CLT page. My patience with this method usually runs out between 20 - 40 pages (i.e. 1000 profiles or less and 30 sec. to a minute of mouse clicks)

(2) Load CLTPlus results into CLTBoss. This means the scraping has been done already, so you can either generate Invelos XML immediately from the online, or run the CLTBoss XML scan if you want the skinny XML output (CLTBoss parses all the dividers, scans for credits and makes a more compact output than Invelos format)

(3) "One-click" totally automatic web page scrape: followed by XML scan. This can work well when your network is fast and unloaded, and you have a small number of CLT pages. Start it up and go play with the dog, your kids or a friend. Come back, and it has either completed perfectly or not. If not, you didn't waste much time. This method usually takes about 1 minute per 10 pages, followed by the XML scan time (same for all methods).

Now, you can also use the automatic method for a large number of profiles / pages. It will either work or not. If you don't mind the possibility of coming back 30 min or 1 hour later to face an error report about missing profiles, then give it a shot, especially if you have a super network, and you have tried it and it seems to work for you. Best ones of these I have run are Tom Cruise (3033), Danny DeVito (2563), Tom Hanks (3887), and Christopher Lee (3516). During the scraping - (1 min per 250 profiles) - you can't mess with the screen. After that, the XML scan runs in a thread like any other Profiler plugin, and you can do other stuff, but of course you will slow things down, or possibly interrupt the network. I say, go watch a movie.

I plan to also do a "Walking Around CLTBoss" video that describes how to use the frilly features.

That's my plan at the moment. Got to get back to the many sidelined home projects before the winter sets in. Then if anybody cares, or if it hasn't been already done, I will start looking for a way to do a solid web page scrape so that is equal to or more reliable than the mouse-click scrape.