Surely that should be clopen-range?

Makes me want to get out my guitar

If you read a string from a port, then it it was UTF–8-encoded by definition. If that port draws from another port with a different encoding — i.e., the port’s job is to convert — then it’s up to the convert port to translate your byte-based consume to a suitable consume on the original port.

But I forgot about bad UTF–8 encodings. If there’s a bad encoding that gets converted to the error character, then string-utf-8-length would not be right.

So, probably string-utf-8-length is not a useful piece for this problem.

About once a month right now

@samth it definitely should not

has anyone made box-plots with the plot library? if so, are there examples of doing this? looking at the candlesticks renderer but that mandates that the x-coordinate is a real

asking in case I’m missing something before I try and write a PR

@alexharsanyi did a very nice talk at Racketfest with fancy plots (and fancy data processing), but I don’t remember if there was any box plot in that demo.

@hazel Haven’t seen box plots for plot.

It would be a good thing that the user computes the extended quartile set himself. Due to the number of different conventions in use of computing the median, and the first and third quartile.

Ah, I see—I misunderstood the way encodings worked, but rereading the “Encodings and Locales” section of the reference, I see it is entirely consistent with what you’re saying. In that case, I think string-utf-8-length is probably good enough.

I can play with the way peeking-input-port behaves, but my main fear is how to handle commits. If I understand correctly, I would need to call port-commit-peeked on the original port, but this would disrupt the location of my peeking port. So if I need to commit part of what I’ve peeked so far, but not all of it, I assume I would have to construct a new peeking-input-port, but perhaps that’s okay. I can experiment.

plot doesn’t really do anything with the data before plotting it, so I’d imagine a PR would just say “where to put the lines”

I’m really surprised that this improved compile time so much. For a big scribble doc, it saved me > 10%. This makes me wonder about the default behaviour, at least when specifying --only --pkgs.