Re: [PATCH v3 4/5] Implement "set cwd" command on GDB

[Pedro Alves <palves at redhat dot com>]

On 09/22/2017 07:56 PM, Eli Zaretskii wrote:
> There's no contradiction between these two sides of this issue.  For
> native debugging, host == target, so the host can expand.  For the
> non-native debugging, you can ask the target to do the expansion and
> store the result.  

You can't, not as the actual value of the setting, because the
setting can be tweaked before GDB is even connected to a target.

I.e., you can do:

(gdb) set cwd ~  # I haven't even connected to a target yet.
                 # Where should this be expanded? 
                 # On host may be incorrect.
                 # '~' -> /home/pedro on this machine
(gdb) target extended-remote foo:9999
(gdb) start
error: /home/pedro does not exist   
# '~' is /mount/home/pedro on 'foo'
(gdb) kill
(gdb) target extended-remote bar:9999
(gdb) start
error: /home/pedro does not exist
# '~' is /nfs/home/palves on 'bar'


It may be useful to display the expanded path as extra info,
like in:

(gdb) set cwd ~foo/bar
(gdb) show cwd
The current directory is ~foo/bar (/home/foo/bar)
                                   ^^^^^^^^^^^^^

though that'd require a new remote protocol packet.

> Either way IMO is better than expanding at run
> time, because the latter makes the expansion dependent on factors
> which could be out of the user control, 

I don't see what the problem is here.

> and also requires every use
> of the value to call gdb_tilde_expand, thus wasting cycles.

I don't think that significant, compared to all the other
work / syscalls / remote protocol roundtrips that we have
to do to start a process.

Thanks,
Pedro Alves