[RFC] Add ARI directory to gdb sources

["Pierre Muller" <pierre dot muller at ics-cnrs dot unistra dot fr>]

  We recently had a discussion about integrating gdb_ari.sh script
into gdb sources directly.

See:
http://sourceware.org/ml/gdb-patches/2012-04/msg00834.html

  Currently the ARI (Awk Regression Index)
script is a separate cvs checkout
and is used to generate three type of ARI reports:

Last release ARI:
http://sourceware.org/gdb/download/ari/
generated normally in the release process.

daily run for current cvs trunk:
http://sourceware.org/gdb/current/ari/

daily run for most recent branch:
http://sourceware.org/gdb/ari/

  The idea behind moving gdb_ari.sh script into gdb sources 
would be:
1) Be able to adapt this script to changes in the gdb sources.
  Example: Currently, there are 14 types of critical bugs
(corresponding to previously eliminated problems that resurfaced).
These are mostly due to some functions that should not generally
be used in the gdb sources (like xasprintf function)
but are present at specific locations:

BEGIN { doc["xasprintf"] = "\
Do not use xasprintf(), instead use xstrprintf"
    fix("xasprintf", "gdb/defs.h", 1)
    fix("xasprintf", "gdb/utils.c", 1)
    category["xasprintf"] = ari_regression
}
/(^|[^_[:alnum:]])xasprintf[[:space:]]*\(/ {
    fail("xasprintf")
}

This means that xasprintf function should appear
only two type in the whole gdb sources,
once in defs.gh and once in utils.c.
  But recent changes (moves of some functions to common
subdirectory), lead to the fact that xasprintf is now
referenced in common/common-utils.h and common/common-utils.c

  This is of course not a real problem, but changing this
in the gdb_ari.sh script would mean that this script would
give wrong information if run on older gdb sources.

2) The second reason was to allow developers or any person
that wants to analyze gdb sources to run this script.


  The question of the location where we should add this script in gdb source
arose.
Joel suggested to put it directly into
gdb directory.
  But after looking at the suggestion, it appears
that you can't just use gdb_ari.sh alone,
the update-web-ari script also uses gdb_find.sh script for instance,
and a modified version of this update-web-air should probably also
be added.

  Thus I would rather suggest to add a new directory called ari.

Suggestions and comments most welcomed,



Pierre Muller
GDB pascal language maintainer
writing as unofficial ARI maitainer