💾 Archived View for splint.rs › banal_code.gmi captured on 2024-05-10 at 10:44:57. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2024-03-21)
-=-=-=-=-=-=-
I can't stand coding helpers which suggest auto-completions, and this is why.
When the spelling checker came along, it helped writing a great deal. Nobody wants the job of looking up how to spell words. Then one dark day, I wrote a twisted sentence, subverting some standard English phrase in order to emphasise a point or make a joke, and blue (not *red*, but *blue*) squiggly lines appeared under my phrase. A right-click informed me that Microsoft or Google or whatever suggested I change the phrase back to the standard. The machine clearly hadn't understood the genius linguistic flourish. I dismissed the prompt with contempt, but I don't think everyone else will dismiss it so quickly. Anyone less secure in their writing could see those blue wiggly lines, and decide to return to the 'proper' (i.e. 'normal', i.e. 'boring') sentence structure.
Style-checkers fix bad writing at the cost of good writing. They make writing 'just okay'.
Code helpers help writing code a lot. I like vim's in-built completion, and the various type-checkers obviously save a lot of headaches. I worry specifically about computers suggesting what kind of function to use, or even suggesting that a function should come next. I think subversive flourishes (clearly written, well-commented, and perfectly documented in triplicate) are worth the cost of some bad code. Actually, I also want to value the bad code - if nothing else, it forces the writer to ask *why* this code feels bad.