Re: [PATCH] PR exp/9103

[Tom Tromey <tromey at redhat dot com>]

Tom> The difficulty is really in deciding the way forward.  I don't know
Tom> what to do.

Just to be clear, here are the issues that I know of.

* The patch assumes that iconv recognizes "wchar_t" as a charset name.
  I have read that this is not universally true, but I have not tested
  this.  I'd be interested in reports on this from folks on non-Linux
  hosts.

* What to do about hosts without an iconv implementation.  There are
  three choices:

  1. Require libiconv on such hosts
  2. Like #1, but provide a simple way to drop libiconv into the src
     tree.
  3. Write a phony iconv-alike that only handles host charset ==
     target charset, probably also providing "gdb_wchar_t" and various
     other macros and whatnot to make it all work.

  Related to this is the question of what to do if we run into a host
  that has a notably bad iconv.

  These are roughly in the order in which I'd prefer to do them... but
  then, the problem does not actually affect any machine I have access
  to.

* I'd like comments on this patch (scroll to the end):

    http://sourceware.org/ml/gdb-patches/2009-01/msg00363.html

  It helps with the C case, and IMO at least creates types more along
  the lines of what I would expect.  But, I'm not an expert here and I
  also don't know whether a similar patch is needed for the stabs
  reader.

Tom