Thanks, Wolfgang. I see I managed to delete the second part of my email before sending it, so here it is:
Suitable characters are a-z, 0-9 (but not initially), and the following: ! $ % & - . @ ^ _ ~ That's the intersection of ASCII, Windows file characters (lowercased). and Scheme identifier characters. This proposal only applies to the scheme and srfi namespaces. ----- Having written that, now I wonder if hyphen is the best separator. Perhaps @ would be a good choice: it's not allowed initially, it wouldn't cost much to disallow it in all other positions, and it stands out, unlike hyphen which is de facto whitespace. On Thu, Jun 5, 2025, 7:50 PM Wolfgang Corcoran-Mathe <[email protected]> wrote: > Forwarding John's message to the list. > > ----- Forwarded message from John Cowan <[email protected]> ----- > > Date: Thu, 5 Jun 2025 18:40:43 -0400 > From: John Cowan <[email protected]> > To: Wolfgang Corcoran-Mathe <[email protected]> > Subject: Re: SRFI 97 library name equivalence? > > I think it was me, but I certainly didn't have any *reason* to do so; it > probably just seemed obvious at the time. > > This name-hyphen-number strategy sounds good to me. > > As for standard characters, pp > > > > On 2025-06-04 21:30 +0200, Daphne Preston-Kendal wrote: > > > On 4 Jun 2025, at 19:36, Wolfgang Corcoran-Mathe <[email protected]> > > wrote: > > > > > > > I'm now aware that (srfi N) is a poor convention & requires too much > > > > memorization of SRFI numbers. I wish I'd paid more attention to > library > > > > naming in the past. > > > > > > It’s worth noting that R7RS small simply states ‘Libraries whose > > > first identifier is srfi are reserved for libraries implementing > > > Scheme Requests for Implementation.’ It says nothing about the > > > structure of this namespace. ... > > > > > > I don’t know who decided to do away with symbolic names for SRFI > > > libraries with R7RS-style names in the first place. John, perhaps? > > > > Sorry, I didn't mean to imply that this convention was in any sense > > standard. > > > > Quite possibly it was John. A quick search suggests that SRFI 111 may > > have been the first SRFI to use the (srfi N) convention explicitly > > (i.e. in the SRFI document, & not just in the sample implementation). > > > > > ... the best convention would probably be > > > (srfi <library name>-<library number>), where the two template parts > > > form a single identifier. > > > > I also like this convention, but it's incompatible with SRFI 261. > > What should we do? > > > > -- > > Wolfgang Corcoran-Mathe <[email protected]> > > > > ----- End forwarded message ----- > > -- > Wolfgang Corcoran-Mathe <[email protected]> >
