Ada and the "start" command

Ada and the "start" command

[Daniel Jacobowitz]

Hi Joel,

Do any of the gdb.ada tests pass for you, using a clean GDB tree?  My
impression is that the "start" command, upon which they all rely, can
not possibly work in FSF GDB.  There's ada_main_name, which correctly
digs the name out of the executable, but it isn't hooked up to
anything.  So "start" always goes to "main".

If the tests can't work in this tree, they should be disabled.  I'm not
sure what the plan for ada_main_name is.

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: Ada and the "start" command

[Daniel Jacobowitz]

On Sun, Mar 06, 2005 at 04:33:37PM -0500, Daniel Jacobowitz wrote:
> Hi Joel,
> 
> Do any of the gdb.ada tests pass for you, using a clean GDB tree?  My
> impression is that the "start" command, upon which they all rely, can
> not possibly work in FSF GDB.  There's ada_main_name, which correctly
> digs the name out of the executable, but it isn't hooked up to
> anything.  So "start" always goes to "main".
> 
> If the tests can't work in this tree, they should be disabled.  I'm not
> sure what the plan for ada_main_name is.

Scratch the last sentence; I found the pending patch on gdb-patches.  I
still do not like tests which will not yet pass being added to the
source tree.

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: Ada and the "start" command

[Joel Brobecker]

> > Do any of the gdb.ada tests pass for you, using a clean GDB tree?  My
> > impression is that the "start" command, upon which they all rely, can
> > not possibly work in FSF GDB.  There's ada_main_name, which correctly
> > digs the name out of the executable, but it isn't hooked up to
> > anything.  So "start" always goes to "main".
> > 
> > If the tests can't work in this tree, they should be disabled.  I'm not
> > sure what the plan for ada_main_name is.
> 
> Scratch the last sentence; I found the pending patch on gdb-patches.  I
> still do not like tests which will not yet pass being added to the
> source tree.

I know what you mean.

At the time the test was added, the test was supposed to be failing for
a short period of time, because it was felt that the problem be fixed
quickly (we had discussed on how we were going to hook that up and
agreed on a plan). Since then, Elena jumped in, approved almost
everything, except one bit. I answered her message, but never received
any response, even after a couple of pings.

We can do two things: Either have somebody else than Elena approve
(or comment) on the patch, or KFAIL the tests. Do you think you can
approve the rest of the patch?

-- 
Joel

Re: Ada and the "start" command

[Daniel Jacobowitz]

On Sun, Mar 06, 2005 at 01:44:21PM -0800, Joel Brobecker wrote:
> > > Do any of the gdb.ada tests pass for you, using a clean GDB tree?  My
> > > impression is that the "start" command, upon which they all rely, can
> > > not possibly work in FSF GDB.  There's ada_main_name, which correctly
> > > digs the name out of the executable, but it isn't hooked up to
> > > anything.  So "start" always goes to "main".
> > > 
> > > If the tests can't work in this tree, they should be disabled.  I'm not
> > > sure what the plan for ada_main_name is.
> > 
> > Scratch the last sentence; I found the pending patch on gdb-patches.  I
> > still do not like tests which will not yet pass being added to the
> > source tree.
> 
> I know what you mean.
> 
> At the time the test was added, the test was supposed to be failing for
> a short period of time, because it was felt that the problem be fixed
> quickly (we had discussed on how we were going to hook that up and
> agreed on a plan). Since then, Elena jumped in, approved almost
> everything, except one bit. I answered her message, but never received
> any response, even after a couple of pings.
> 
> We can do two things: Either have somebody else than Elena approve
> (or comment) on the patch, or KFAIL the tests. Do you think you can
> approve the rest of the patch?

Or we could politely ask Elena again, and KFAIL the tests - that's my
prefered solution.

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: Ada and the "start" command

[Joel Brobecker]

> Or we could politely ask Elena again, and KFAIL the tests - that's my
> prefered solution.

Sure, I will do that shortly.

-- 
Joel