CLOSED: Inexplicable CHS find/search results - rules not understood.

[IainB]

Interesting behaviour in CHS:
Attempting to locate in the Grid the exact clip that was being displayed in the Memo panel - CHS does not always display that clip highlighted in the visible Grid pane, so you have to scroll around for it - I copied a partial string from the Title field of the clip in the Memo panel.
The string was "Corporation - psychopathy".

I then pasted that string into the Find/Search box (where the mouseover info says "filter clips using these strings"), and the clip disappeared from the Memo pane, and the Grid was empty. That is, Search could find no clip record in the database that had that string - which was clearly not the case as the string was copied from an existing clip.

However, I had previously thought that CHS Search would unfailingly and almost instantaneously find what you asked it to search for, so I presumed that the string had something in it that was causing Search to be "blind".
Thinking that maybe it was the dash/minus character in the string ("-") that could be causing the problem, I experimented with the string. The results (hits) from CHS were interesting, though somewhat confusing:

• "Corporation - psychopathy" - CHS fails to find it (no hits), even though that is the exact string as copied from the Memo pane.
• "Corporation psychopathy" - (with just 1 space or n spaces between the two words) CHS hits show that it consistently finds the clip(s) with those words in.
• "Corporation -psychopathy" - CHS had several hits, which seemed to be only those clips with "corporation" in it, but I'm not sure about the consistency. I suspect that there may well be more to it, as I didn't go through all the hits.
• "-psychopathy" - CHS had lots of hits, but I couldn't figure out what was a consistent reason (there will be one) for selecting those hits. However, it didn't seem to be the word "psychopathy" anyway.
• "Corporation- psychopathy" - CHS had no hits.

What this would seem to indicate is that, whilst we might take it as a given (or I had, anyway) that CHS will consistently find clips in its database, and thus we might rely on CHS to do just that, the reality could be somewhat different.
I am now unsure of the rules governing this CHS search function.

[mouser]

You approached this as a detective and you got very close to figuring out what was happening.

The - is treated by CHS search as a NOT.

I think you'll find this explains the combination of results you got.

For example when you were typing "corporation -psychotherapy" it was finding all clips with corporation in them but NOT psycotherapy.

Now i guess what i do need to add is some way to escape that - so you can search for it normally.

[mouser]

Another thing I could add is an option that would remove this special behavior of the - character.

[IainB]

Ahhh, I see. Thankyou.
Well, that's the rule I was unaware of! That is, "-" immediately(?) preceding a character/string means "NOT".
Sorry if I should have been aware of that. I did look in the Help file for enlightenment, but didn't spot it if it was there.

I figured it would have to be consistent, whatever it was - and it is.
I think it's a great rule to have - very smart search.   
So don't change it, please!
Is this (and other rules) documented somewhere? I need to know...
For example, how close to the negated search string does the "-" have to be?
Presumably "Corporation - psychopathy" meant "NOT space character" - right?
 
CHS search otherwise does very well - and more than you'd expect. For example, if I search for "Fallout" the hits include "Fallout3", which is exactly what I might be looking for.

[IainB]

...Presumably "Corporation - psychopathy" meant "NOT space character" - right?...

-IainB (March 14, 2016, 09:45 PM)

Hmm. If that is the case, then which characters are being taken as delimiters between the contiguous sub-strings in the string "Corporation - psychopathy", or is space the delimiter, so that the embedded "-" is not directly related to a string and is thus taken to negate the entire string?   
Confuzzled.

[mouser]

my memory is spaces as delimiters.  if you have just a - by itself it will probably be ignored.

[IainB]

my memory is spaces as delimiters.  if you have just a - by itself it will probably be ignored.
_________________________

-mouser (March 15, 2016, 11:10 AM)

Oh! Then that means:

1. "Corporation - psychopathy" - CHS fails to find it (no hits), even though that is the exact string as copied from the Memo pane.

- cannot be explained.

[mouser]

true, ill have to look into that case.

[mouser]

Searching for - by itself should work properly in next release.

[IainB]

Feedback on this Beta: I can confirm that CHS v2.36.0 Beta search:

• finds a minus symbol or small dash ("-") with no problem.
• treats an immediately preceding dash (which itself has a preceding space) as a NOT in the clip search - e.g., the search string "corporation -civil" will find all occurrences of clips that contain "corporation" and which do NOT contain "civil".
• correctly searches for embedded dashes - e.g., in hyphenated words such as "corporation-civil".

Very nice, thanks.

Thought I should probably mark this thread as "closed" now.

[mouser]

glad to hear it, thanks Iain.