Neil Van Dyke writes:

> I think taxonomies for classifying libraries for human location are
> (or should be) more ephemeral than the universal identifiers of the
> libraries themselves.

I think you're right about that.

The only disadvantage I can think of for having a flatter structure is
that perhaps there might be more conflicts with people wanting the
same name for their library.

If the name of the Spiffy web server is "web:server:spiffy", then
someone else can have a Spiffy graphics drawing library if they want
to.

I agree with you that a more hierarchical or taxonomy based naming
scheme will probably result in a greater number of name changes over
time.

I expect some amount of library name changes is inevitable.

Library name changes might not be a problem to the extent we can avoid
reusing the old name.

If I decide to change the name of my library, perhaps I could release
a version of the library under the new name and a version under the
old name which forwards imports to the new library name:

    ;; forward old to new

    (library andrewwilcox.name::neato r6rs
      (import new-place::neato)
      (export a b c d e f ...))

If something like this would be a workable solution, and library name
changes were not a big deal, then perhaps the cost of an increased
number of library names changes due to using a hierarchical taxonomic
structure would be ameliorated.

Andrew