[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: whoa

This page is part of the web mail archives of SRFI 103 from before July 7th, 2015. The new archives for SRFI 103 contain all messages, not just those from before July 7th, 2015.

Taylor Campbell wrote:

> I'm not keen on what has become the conventional mapping from library
> to nested directory structures in implementations of the R6RS, or for
> that matter with the R6RS's library system at all, but I have to say
> I'm pretty astonished by the immensity of this SRFI.

I don't think it's immense.  I don't think it's small either.  It's size
is proportional to the features and description I wanted it to have.
Most of the features in it are already available from R6RS systems.  It
standardizes them and addresses known issues with them, and it adds
qualities I think will be useful for building upon to programmatically
manage collections of library files.

I have a couple sizable collections of libraries, and I review others'
sizable collections.  I don't think I could have as effectively
organized and managed my collections, or as easily reviewed others',
without the compound library names and convention for hierarchical
organization of files.  I want to make and review more, possibly large,
collections, and I want automation tools for managing collections, and
this SRFI has a design I think I'd like to have.  I think this SRFI can
help promote creation and sharing of larger numbers of larger

> I expected, perhaps, a sentence such as the following to explain the
> entire specification, modulo details of term definitions:
> `A Scheme system keeps a list of directory pathnames called its search
> path, relative to which it looks in order for libraries' source code
> by mapping their component names to file names, mapping alphanumerics
> and hyphens to themselves, and mapping all other code points by
> percent-encoding.'
> Maybe there need to be a couple more sentences about an environment
> variable storing the search path, putting version numbers before the
> `.sls' suffix 

If it were that small, it would lack features and qualities I want it to

> (which is silly enough as it is -- what was wrong with
> `.scm', again?), 

I, and I think others, like `.sls' because it distinguishes files
containing R6RS-like libraries from all other file types.  A distinct
extension allows identifying library files versus non-library files by
just the file name, without having to look in the file.  I use `.sps'
for R6RS programs.  I still use `.scm' for non-R6RS files (i.e. files
which are not R6RS-specific, such as R[45]RS code I run as R6RS code).

> and bumping the directory used to find a particular
> library up to the front of the search path when finding its imports so
> that it gets its neighbours if possible.

I don't think I'd like that.

> Anything more than that, I can't keep in my head when trying just to
> write Scheme code!

I don't think using it is as head-consuming as you're thinking.  You
probably have to become familiar with it, like any system.

: Derick