[HN Gopher] Horrible edge cases to consider when dealing with mu...
       ___________________________________________________________________
        
       Horrible edge cases to consider when dealing with music (2022)
        
       Author : pabs3
       Score  : 54 points
       Date   : 2023-11-11 11:33 UTC (11 hours ago)
        
 (HTM) web link (dustri.org)
 (TXT) w3m dump (dustri.org)
        
       | jdpigeon wrote:
       | shudder
        
       | slater- wrote:
       | yes. we should resist the tech industry's relentless binary
       | categorization of everything. "oh no! i couldnt find it in the
       | dropdown! what ever shall i do?"
        
       | euroderf wrote:
       | > GZR, also known as g//z/r on their first album, geezer on their
       | second, and GZR on their third.
       | 
       | I would be impressed if they had changed the band name from album
       | to album BUT all three albums had the same identical album name.
        
         | morelisp wrote:
         | Wow if only we had an album_grouping_key or musicbrainz_albumid
         | or something we could have solved this 20 years ago.
         | "Horrible."
        
           | jwells89 wrote:
           | Assuming a music library use case, one could address it with
           | UI alone (making the the method of grouping under the hood an
           | implementation detail) with "merge artist" and "merge album"
           | functions surfaced in places like the context menu that
           | appears when right-clicking an artist/album.
        
       | pwdisswordfishc wrote:
       | title += " publishing metadata"
       | 
       | I was hoping for something about microtonality and non-equal-
       | temperament scales.
        
         | kbenson wrote:
         | Yeah, as someone that knows liyyle to nothing about music
         | theory, I was hoping to learn something interesting along those
         | lines, even if I admit it's unlikely to be useful to me.
         | Instead it's about storage schemas and correctly mapping domain
         | values.
         | 
         | Which, honestly, is more useful to me, but also I find it much
         | less interesting because it's well trod ground.
        
       | specproc wrote:
       | I really enjoyed reading that, the author clearly has taste.
       | 
       | Reminds me of the hours I'd spend as a kid trying to work out
       | folder structures for my MP3s.
        
       | Joeboy wrote:
       | > Some tracks are long
       | 
       | I guess it's only talking about recorded tracks, but 13 hours is
       | rookie numbers[0]
       | 
       | [0]
       | https://en.wikipedia.org/wiki/As_Slow_as_Possible#Halberstad...
        
       | morelisp wrote:
       | > Some artists will benchmark your utf8 support:
       | 
       | Sorry but until I see the band that varies their album names only
       | in normalization scheme this is baby shit.
        
       | Karellen wrote:
       | aka "Falsehoods programmers believe about recorded music" :-)
        
       | sneak wrote:
       | alt-j is a bad transliteration of the actual band name which is
       | what you get from option-j on a mac (a triangle iirc).
        
       | jrockway wrote:
       | None of this seems too terrible to handle. The title of the song
       | can be an empty string, so use *string/Maybe string instead of
       | string to distinguish between unset and "some idiot made the song
       | literally untitled". I am surprised there are no songs that use
       | characters only available in two different character sets, or are
       | the same Chinese character repeated with multiple different
       | writings (Simplified Chinese, Traditional Chinese, Korean,
       | Japanese).
       | 
       | I suppose that some people are running into trouble by using
       | CHARACTER VARYING(n) instead of TEXT in their database, but if
       | you pick the most likely database for your local music player,
       | then it doesn't matter. (sqlite doesn't care, basically.)
       | 
       | The song that's named after a virus test signature is amusing. I
       | guess encrypt your filenames and database so that the OS can't
       | pry into the details. Since you hopefully didn't write your music
       | player in C, there should be no way that a virus can affect your
       | music player.
       | 
       | (While I was researching this post, I found an interesting Defcon
       | talk about what happens when you turn the EICAR virus signature
       | into a QR code and scan it.
       | https://www.youtube.com/watch?v=cIcbAMO6sxo A lot of stuff
       | breaks! I almost got forced into adding similar breakage. At work
       | we make what is essentially a database, and we got acquired by A
       | Large Company, which had to do a security audit of our software.
       | One of the "security vulnerabilities" in our product was that we
       | don't scan the data in the database for viruses. We pushed back
       | to not do this, because one of our customers literally uses our
       | system to store viruses they analyze. But, it appears that a lot
       | of other teams got this same suggestion and just made their price
       | checker or passport validator crash when the barcode resolves to
       | a virus. Easier than input validation or not calling
       | `system(whatever the barcode resolved to)` I guess?)
        
       | NeoTar wrote:
       | I think he missed the band called just "A". Who released the song
       | "nothing" https://en.wikipedia.org/wiki/Nothing_(A_song)
       | 
       | I can recall having trouble searching for their album on Amazon
       | in the day, and indeed it doesn't seem any easier today.
        
       | crtified wrote:
       | "Convoluted esoteric language which requires all manner of
       | special measures to even read!", ...said the computer
       | programmers.
        
       ___________________________________________________________________
       (page generated 2023-11-11 23:00 UTC)