assuming you're a new free sw project, and you wrangle with finding a name for it. should there be a blacklist of forbidden names of which your projects name should be of a certain minimum levenshtein distance? like is starlink too close to stalin?

@stf does a distance of 2 imply there can only be 26 different 3-letter projects?

@stf Given that 1) naming things is hard and 2) you shouldn't reinvent the wheel, you just take the name of a similar project. (and this is how eventually every OSS project becomes "curl")

@supersingular ackshully levenshtein distance allows adding and removing letters.

@buherator i think another good name to aspire to is "sphinx"

@stf aaackchually for fixed length strings a levenshtein distance of 1 is equivalent to a hamming distance of 1, and it follows that levenshtein distance >= 2 iff hamming distance >= 2.

but my initial guess of 26 was too low - using the singleton bound there is a set of 26^2 length-3 strings with a minimum levenshtein distance of 2.

though this doesn't generalize past a minimum distance of 2: https://math.stackexchange.com/questions/5091079/bounds-on-the-number-of-strings-of-fixed-length-that-have-levenshtein-distance

@supersingular aaaaaaackshully a hamming distance of letters is often very surprising, e.g. the hamming distance of 'c' to 'd' is actually 3, since hamming distance is measured in bits, and 'c' == '0b1100011', while 'd' == '0b1100100' - assuming ascii encoding of the letters.

shit i'm wrong, hamming distance also requires a list symbols, so if it is letters and not bits, you are correct, and i learned something.

@stf so I was curious how limiting this would be - and here's a list of the conflicts for the packages in nixpkgs: https://github.com/ambiso/optimal-code-levenshtein/blob/main/conflicts.txt

i.e. these are all pairs of package names that have a levenshtein distance of < 2.