Some comments on procedure names Lassi Kortela (07 Oct 2019 15:18 UTC)
Re: Some comments on procedure names Linus Björnstam (08 Oct 2019 19:39 UTC)

Re: Some comments on procedure names Linus Björnstam 08 Oct 2019 19:39 UTC

On Tue, 8 Oct 2019, at 19:52, Linus Björnstam wrote:
>
> On Mon, 7 Oct 2019, at 17:18, Lassi Kortela wrote:
> > - tflatten -- With flattening procedures, there's always the question of
> > whether it flattens only one layer of lists, or all layers recursively.
> > This looks like it flattens all layers. I don't know if there are
> > standard names for these variants; I've used "flatten-once" and
> > "flatten-all".

Since all schemes that have srfi-1 has concatenate, I think flatten is
a good name in contrast to that. Racket uses flatten, and apart from
that I find very little info about other names used.

> >
> > - tsegment -- This operation is variously known as 'segment', 'group',
> > etc. Is one of those names preferred in Scheme as a whole?

Tsegment and tpartition used to be called tpartition-all and tpartition
respectively (names stolen from Closure). Names were changed after a
suggestion by John. Naming them in a clear and easy-to-separate way is
nearly impossible, but this way separates the meaning of tsegment from
partition of srfi-1

> >
> > - tadd-between -- This operation is variously known as 'interpose',
> > 'delimit', etc. Is one of those names preferred in Scheme? If not, the
> > obviousness of "add-between" is nice :)

 interpose is the clojure name. As someone with another mother tongue
 than english, I had no idea what interpose meant. John suggested
 add-between which I liked and has some precedence in racket.

> > - tconcatenate -- Could this be 'append' for symmetry with tappend-map,
> > or is it deliberately called concatenate instead of append for a
> > particular reason? SRFI 1 has both append and concatenate; append is the
> > usual one featured in most Scheme programs so it's more familiar.

 concatenate and append do slightly different things, and I think the
 workings of tconcatenate is closer to concatenate than append, at least
 when used with xxx-transduce (which is probably the best use-case for
 it, since a streaming use is rather useless).

> >
> > - tdelete-neighbour-dupes -- Would the American English spelling
> > "neighbor" be OK? Almost all Scheme code uses AE spelling.

Good catch. British english is what I was taught in school, but that
has been diluted through long exposure to american TV. Even though I
prefer the British spelling I agree with this change.
>
>
> >
> > - revery -- The identity value is not stated in the SRFI. According to
> > the implementation, the identity if #t, which is the usual one and
> > matches SRFI 1.
>
Will add! Good catch.
>
> >
> > P.S. Linus, congratulations on your first SRFI :) Very good work!
> >
>
Thank you. John asked me to submit a much more controversial one (based
on a syntax-case implementation of rackets for loops for guile). It
will be interesting to see whether I can get that through first draft
status :D

I will try to answer the other emails tonight, but the fatherly duties
come first! Thank you for your input!

Best regards
Linus