Grouping is Wrong with a Length Tolerance=1

[therube]

Grouping is Wrong, sometimes, with length_dupe_tolerance<>0.

(i have not seen the issue when tolerance=0)

1. the two 2.11.03 should be Grouped (tolerance=1)
2. setting (tolerance=0), the 2.11.03 files are no longer there
.

Everything - Grouping Wrong, Find Length Dupes, sometimes2.png (49.8 KiB) Viewed 1065 times

[void]

length_dupe_tolerance is now specified in milliseconds (instead of seconds).

This change was made in Everything 1.5.0.1373a.



Please try setting length_dupe_tolerance to: 1000 for one second.

[therube]

I'm guessing that Grouping is not taking Tolerance into consideration?

length_dupe_tolerance=1000
expecting to find files of Len 1.23.00.

1. filtered search, & my 2 expected files now do turn up, though they are not Grouped together.
2. all 1.23.00 Lengths, none seemingly are exactly the same Length, so all are Grouped individually.
(Other searches do show some same Length files Grouped together, though I don't have the physical files so can't confirm their Length "identicalness" [heh].)
.

Everything - Length Tolerance 1000.png (50.05 KiB) Viewed 1034 times

Not sure yet how this will work out in practice, but I'll give it look.

[void]

Thank you for clarifying.

Everything 1.5.0.1374a will now apply your length dupe tolerance to the group lines.

[therube]

1374.

(With a quick look...)


Grouping now looks good.
(Actually I'm sure it will be FAR better then 1373.)

But (& I haven't thought about it yet), Coloring within the Group is off.

1. Length duplicates, coloring is off
2. Size duplicates, coloring is consistent within the Group
.

Everything - 1374 Grouping now looks good, Coloring is off 2.png (29.94 KiB) Viewed 979 times

[therube]

I've got to think more about this & how much tolerance I can tolerate .

What potentially happens at some point, is the tolerance factor causes an "endless" linkage of tolerant files, resulting in huge groupings.

Like I came up with a group starting at a (Length) 1.05.00 & by virtue of having a 900ms tolerance, it picked up 1.04.59, which picked up 1.04.58, which picked up 1.04.57 ... which picked up 1.04.43, before the group finally broke.

So my tolerance of 900ms ended up grouping files of 16 seconds (worth of tolerances).

Now this was on an unfiltered search.
Once I do entered filtered searches, thinking the results would be more appropriate.

And on my end, thinking I'll also end up with better results by lowering the tolerance too.
I'll have to mess with it, maybe dropping to 450ms... & see how that works out.

[void]

Like I came up with a group starting at a (Length) 1.05.00 & by virtue of having a 900ms tolerance, it picked up 1.04.59, which picked up 1.04.58, which picked up 1.04.57 ... which picked up 1.04.43, before the group finally broke.

This is expected.
The difference between each item will be 900ms or less.
However, the total difference for the whole-group may be larger than 900ms.



Everything 1.5.0.1375a will now calculate the group color from the length / length_dupe_tolerance.
This means if you set length_dupe_tolerance to 1000, then each unique second will have its own color.
While not correct, it should be an improvement over the current display.
Does this help?

[therube]

1374

yes, with Filtered Search, Grouping works out better
(& even though there still are "collisions", they're
at least decipherable [in my case... so far, at least]

haven't changed Tolerance (yet), still 900ms

Coloring, once that is working better, then this entire
group-tolerance-color deal will be much smoother)

---

1375 (quick look)

Looks like coloring is working out better.
Most often, coloring now is more apt to point out "pairs" in a group.

1374 vs. 1375
.

Everything - Coloring 1374 vs 1375.png (18.86 KiB) Viewed 919 times

.
4.15.26 are the "same" & should be grouped & colored alike
3.59.26 & 3.54.47/6 are actually totally unrelated, but the like coloring in 1375 still helps to confirm the Grouping