ansi-tests failures, pretty-printer

[Bruno H. <br...@cl...>]

Hi,

I've revisited the ca. 200 new test failures of the ansi-tests testsuite.
For those that now are flagged as "unexpected failures" I have a fix in
the queue.

For the following, could you, Sam, please take a look?

  NAMESTRING.3 NAMESTRING.4 PATHNAME-MATCH-P.4
  PARSE-NAMESTRING.4 ; only the second value can be fixed
  DIRECTORY.6 DIRECTORY.7
  PPRINT-TABULAR.6

And for the many failing PPRINT-* tests, I think it requires a rewrite
of parts of the pretty printer. It's not just an isolated failure.

Sam writes in the TODO:
> Many pretty-printer test failures.
> (Sam - but Bruno has to fix the risky-test in iofkts.tst first)

Took a look at it, and found that I understand nothing about it. I think,
it will take me 2 or 3 months to work on the pretty-printer. Probably I will
start by writing the ANSI CL functionality in Lisp, not C, (without
*print-circle* handling), and then move selected parts from Lisp to C. How
it will integrate with the existing pretty printer, I have no idea.

This is not something I would like to hold up the MOP release for. (The
pretty-printer stuff is not a regression, right? It's in the same state as
in the 2.33 release?) So my proposal is to schedule this for the 2.35
release, not 2.34.

Bruno