5 ms·
https://en.wikipedia.org/wiki/RAS_syndrome https://en.wikipedia.org/wiki/RAS_syndrome
by yewenjie 5y ago
https://en.wikipedia.org/wiki/RAS_syndrome https://en.wikipedia.org/wiki/RAS_syndrome
- PaulDavisThe1st 5y agoNo. This is not RAS. It's a continuation of a long trend in *nix/FLOSS for using recursive acronyms. One of the earliest: GNU: GNU's Not Unix The one closest to me (though I didn't invent it): JACK - the JACK Audio Connection kit.
- mumblemumble 5y agoIt is. See the difference: GNU Operating System -> GNU's Not Unix Operating System Fish Shell -> Friendly Interactive Shell Shell One expands to include the original acronym, but doesn't duplicate the descriptive extra words. The other doesn't expand to include the original acronym, but does duplicate the descriptive extra word. In other words, one is recursive but not redundant, the other is redundant but not recursive.
- trogdor3000 5y agoWhat about bash shell? Bourne Again SHell shell. That's the same thing. But i think it can be a helpful distinction between using bash shell as an interactive shell and using the bash scripting language.
- mumblemumble 5y agoYup. I'm one of those folks who thinks that RAS syndrome is an entertaining and occasionally useful language quirk, and not at all a problem. For example, even within a computing context, saying "Fish shell" just might help someone understand that you're talking about the Unix shell and not the cipher, and would always help someone who's not familiar with Fish to at least understand that you're talking about an alternative Unix shell. By contrast, whenever someone posts a Hacker News article about ML, the first comment is invariably someone complaining that they thought it would be about machine learning, and this is unnecessarily confusing, and a borderline injustice, it should change its name, etc. It would probably be easier for everyone involved if it were idiomatic to say "ML language", even thought doing so would be redundant.
- CRConrad 5y agoBut ML preceded ML (at least as a somewhat widespread topic of discussion), so it's the machine learning fans who get to come up with alternative terminology.