Enhancement: Allow element weights

[walles]

Based on walles/riff#14.

I have two string collections I want to examine.

There are two equally valid LCS solutions.

In this case, I prefer one of them, but Diffus produces the other.

To handle this, I'd like to put weights on the strings in my collection, so I can tell Diffus:

• Treating string black as a change is expensive
• Treating string as a change is cheap
• Treating string \n as a change is cheap

Then, in the exact same case, Diffus would prefer the same LCS solution that I do.

[Jim-Holmstroem]

Hi @walles , riff looks really nice :D

Thought about this for a bit
I think the issue will be either you'll have to implement weighted LCS approximation (which seems to be rather advanced to get a decent complexity and running speed) or alternative 2 shoe horn something that will most likely only work for simple cases and might make it more cumbersome for a lot of other use cases

But I might be wrong, if you have deeper insight into this I'm all ears :)

[walles]

If I was good at LCS I would have provided a patch :).

Some very superficial research tells me what I'm requesting is called "Heaviest Common Substring".

No matter what it's called though, I don't see myself implementing this any time soon.

Feel free to close, or to keep open for documentary (can I use "documentary" like that?) purposes!

[Jim-Holmstroem]

its a really nice idea in general

I'll keep it in the back off my head in case something presents itself

for future reference: https://www.sciencedirect.com/science/article/pii/S157086671000002X

(regarding "documentary", not a native English speaker myself, but I usually use "for future reference".