@samth @robby @lexi.lambda I think there’s an issue with the srfi/19 implementation. The locale bundles are only loaded once when the module is required. Therefore if one changes the locale during run, the dates keep being translated in the locale that was set on load.

Happens here: https://github.com/racket/srfi/blob/master/srfi-lite-lib/srfi/19/time.rkt#L101

@ben redex-match leaves the duplicates in so you can debug to find the exponential slowdowns that they cause

Other tools remove the duplicates (like meta functions)

How often are snapshots being built?

is it nightly or more often?

@githree nightly

@jerome.martin.dev is that part of the srfi 19 specification?

thanks @samth, do you think there is a chance this commit (from 11h ago) was included in the last snapshot build: https://github.com/racket/racket/commit/d3aa7e90e7c27cfbf6adad8db2e2c3ba6abac6a7

Hard to be sure

@samth Well SRFI/29 describes how locales are handled, while SRFI/19 describes how to handle dates. But in SRFI/19 it is specified in every date formatting procedure that the appropriate format should be selected from the current locale

They don’t specify how to handle locales, but they specify that locales should be used when printing dates.

Which is buggy in our racket implementation right now, because it only uses the first locale at load time, then never bother to check again if the locale changes.

which gives strange results like that: (require srfi/19 srfi/29) (parameterize ([current-language 'es]) (date->string (make-date 0 0 0 0 30 3 2018 0) "~A, ~B ~e, ~Y")) "Friday, March 30, 2018"

while at the same time: (current-locale "es") (require srfi/19 srfi/29) (date->string (make-date 0 0 0 0 30 3 2018 0) "~A, ~B ~e, ~Y") "viernes, Marzo 30, 2018"

I can do a pull request if you want. I justed wanted to make sure it was a bug.

I just did a git pull and rebuild, but encountered an error: make[11]: *** No rule to make target `.../racket7/racket/collects/syntax/parse/experimental/dset.rkt', needed by `../GRacket3m.app/Contents/Info.plist'. Stop.

do I need to update anything?

update: I used a clean checkout and it worked

@samth @robby @lexi.lambda I wrote an issue to discuss the implementation of SRFI/19 before I start working on a fix. Can you tell me what you think? https://github.com/racket/srfi/issues/6

I don’t have any strong opinions here (or expertise).

What you wrote in the PR makes sense to me tho

I know nothing about srfi/19, really… I’ve used gregor instead ever since it was released.

Has anyone else using codeblock-pict noticed that the lines can have varying heights when using unicode characters?

(Including lambda. I’m also using the acmart monospace font.)

Okay, I have an idea. Does anyone know if its possible to merge two fonts together in Racket?

Like, have it use font A, but for any missing glyphs in font A use font B.

Ok thanks, I’ll do that change then :slightly_smiling_face:

Yeah sorry. I contacted the three of you ’cause you were on the contributer list on github but I guess there’s better to do than remember every srfi by heart x)