[asdf-devel] Quick note on tests

Robert Goldman rpgoldman at sift.info
Thu Sep 13 15:33:52 UTC 2012

On 9/13/12 Sep 13 -9:37 AM, Robert Goldman wrote:
> Running the tests on the latest git version on Mac OS X Lion, CCL
> 64-bit, version 1.8, I got four failing tests:
> failing test(s):  test-configuration.script test-multiple.script
> test-static-and-serial.script test-utilities.script
> Haven't had a chance to investigate further -- I suppose this could be a
> local issue -- but thought I'd pass it on sooner, rather than later.
> More as I have details.
> cheers,
> r

OK, seems like exiling functions to ASDF-UTILS and no longer exporting
from ASDF is causing this failure.

Question: what's the removal of these exports doing, aside from
assuaging a feeling of tidiness?  I get it that one might not like
people using ASDF as a source of utilities, but is that not-liking
strong enough to break existing code (or in this case, re-breaking
code)?  [Faré, your ILC 2012 stuff looks like a lot more exciting place
to spend hours than tidying ASDF!]

I also get it that Xach doesn't like use of ASDF as a utility source for
Quicklisp, because it introduces a dependency on bleeding edge versions
of ASDF.  But that doesn't seem to me to be a problem with use of ASDF
for utilities per se --- the problem instead is that ASDF doesn't have a
stable API[*].  Maybe instead of refusing to export these utilities, we
should do a feature freeze, and defer future API changes (any changes to
exported symbols) to ASDF 3, which may or may not arrive before the
coming of the programmer's choice of Messiah.

Freezing the API would also have the effect of encouraging the
development of alternative sources of utility functions.  As people want
more functionality, they will be driven to develop it in a freestanding
library, and that will naturally motivate them to move away from using
ASDF as a utility source in favor of a new library that does not have
ASDF's special status.  [As I read it, one of Xach's objections to the
use of ASDF as a utility source is that it has a special status.  This
seems an eminently reasonable ground for objection.]


* There is a separate problem that people might like to see ASDF wither
away, but I'd be happier to postpone preparations for said withering
until it is sooner to come.  The better is the enemy of the good, and
all that.

More information about the asdf-devel mailing list