MI output change

MI output change

[Andrew Cagney]

Hello,

The MI output is currently littered with:

(gdb)
PASS: gdb.mi/mi-var-cmd.exp: create lsimple->integer
-var-create int * int
&"Attempt to use a type name as an expression.\n"
&"mi_cmd_var_create: unable to create variable object\n"
^error,msg="mi_cmd_var_create: unable to create variable object"
(gdb)

(note how the error appears as both the result and as standard out)

My recent exception handling rewrite means that this can finally be 
fixed.  However that means that the MI output has technically changed - 
no one should be relying on the message appearing on stdout but hey ...

Is this going to be an issue?

Andrew

Re: MI output change

[Bob Rossi]

On Fri, Jan 14, 2005 at 03:54:07PM -0500, Andrew Cagney wrote:
> Hello,
> 
> The MI output is currently littered with:
> 
> (gdb)
> PASS: gdb.mi/mi-var-cmd.exp: create lsimple->integer
> -var-create int * int
> &"Attempt to use a type name as an expression.\n"
> &"mi_cmd_var_create: unable to create variable object\n"
> ^error,msg="mi_cmd_var_create: unable to create variable object"
> (gdb)
> 
> (note how the error appears as both the result and as standard out)
> 
> My recent exception handling rewrite means that this can finally be 
> fixed.  However that means that the MI output has technically changed - 
> no one should be relying on the message appearing on stdout but hey ...
> 
> Is this going to be an issue?

Seems like a nice improvement. 

The message in "msg" does not have a newline, while the one after '&'
does. Does this matter at all?

Bob Rossi

Re: MI output change

[Andrew Cagney]

Andrew Cagney wrote:
Two ya's, no Na's.  All but 'done' :-/
> Hello,
> 
> The MI output is currently littered with:
> 
> (gdb)
> PASS: gdb.mi/mi-var-cmd.exp: create lsimple->integer
> -var-create int * int
> &"Attempt to use a type name as an expression.\n"
> &"mi_cmd_var_create: unable to create variable object\n"
> ^error,msg="mi_cmd_var_create: unable to create variable object"
> (gdb)
> 
> (note how the error appears as both the result and as standard out)
> 
> My recent exception handling rewrite means that this can finally be 
> fixed.  However that means that the MI output has technically changed - 
> no one should be relying on the message appearing on stdout but hey ...
> 
> Is this going to be an issue?
> 
> Andrew
> 

Re: MI output change

[Bob Rossi]

On Tue, Feb 08, 2005 at 11:21:46AM -0500, Andrew Cagney wrote:
> Andrew Cagney wrote:
> Two ya's, no Na's.  All but 'done' :-/
> >Hello,

Has this patch been committed yet? The last failing MI testcase I have
in the syntax checked MI testsuite is one related to this problem.

If you've committed the change for this, I still have an example that
behaves as below.

Bob Rossi

> >The MI output is currently littered with:
> >
> >(gdb)
> >PASS: gdb.mi/mi-var-cmd.exp: create lsimple->integer
> >-var-create int * int
> >&"Attempt to use a type name as an expression.\n"
> >&"mi_cmd_var_create: unable to create variable object\n"
> >^error,msg="mi_cmd_var_create: unable to create variable object"
> >(gdb)
> >
> >(note how the error appears as both the result and as standard out)
> >
> >My recent exception handling rewrite means that this can finally be 
> >fixed.  However that means that the MI output has technically changed - 
> >no one should be relying on the message appearing on stdout but hey ...
> >
> >Is this going to be an issue?
> >
> >Andrew
> >