Re: New directory src/gdb/misc?

[Andrew Cagney <ac131313 at cygnus dot com>]

> Eli Zaretskii writes:
>  > > On Thu, 7 Mar 2002, Daniel Jacobowitz wrote:
>  > > > > This leaves ``gdb/common'' and, (new suggestion) gdb/misc/ as possible 
>  > > > directory names.
>  > > > > 'misc' doesn't convey any information.  It's like calling something
>  > > 'new' in its documentation - not useful for very long.  I'd vote for
>  > > common.
>  > > Either `common' or maybe `lib' (or `gdblib').  I agree that `misc' is not 
>  > such a good idea.
> 
> how about 'generic' (stolen from gdbtk) or 'shared' or 'general' ?
> 
> Elena

`Generic' could be interpreted to mean everything except the 
architecture stuff.

(I think I regret raising this one :-)

Starting again ...

Perhaphs I'm trying to simultaneously squeese a square and triangular 
peg through a round hole

- ``nat'' stuff related to host=target.  gdbserver and gdb have the 
potential to share this.  Signal mapping and ptrace() are probably the 
only things that live here.  Shared libraries, threads, core files, ... 
are all no longer ``native''.

- ``utils'' for just getting gdb to build.  safe_strerror(), xfree(), ...

This does open up the more general question of how to restructure GDB. 
Hopefully that discussion can be side stepped and just persue this to 
the point of being fairly sure we're not heading in a completly wrong 
direction.

Is the right direction to split things across ``functional'' (utils, 
nat, isa/abi, ..) lines rather than ``usage'' (common to gdb/gdbserver, 
commont to gdb/sim, ...) lines?  A functional split would result in 
gdbserver cherry picking from random directories the bits it wants from 
core gdb.

Andrew