GDB 7.10 release 2015-08-21 status update

GDB 7.10 release 2015-08-21 status update

[Joel Brobecker]

Hello everyone,

I think we are definitely getting closer to release time!

The one question that might still be open is whether there is still
work to be done regarding the "target:" default sysroot.

As far as I know, there has been some awesome work between Pedro and
Gary to significantly reduce the time it takes to perform the file
transfers.

There was also the question of being able to interrupt an on-going
file tranfer, and I saw some promising patches, although Sandra reported
that the quick-and-dirty proof-of-concept left GDB in an inconsistent
state after interruption...

There was also a proposal to revert the default sysroot value back
to "", with some added logic to make the "target:" implicit in some
situations; but, as far as I know, this has been abandoned (and
looking back, I would agree).

So, what do we still need to do before we can release?

  - Finish the work on being able to interrupt the file transfer?

  - There were some discussions about displaying a warning,
    and I don't see anything in the current sources; are we still
    thinking about that?

Anything else?

Thank you!
-- 
Joel

Re: GDB 7.10 release 2015-08-21 status update

[Pedro Alves]

On 08/21/2015 01:26 PM, Joel Brobecker wrote:
> Hello everyone,
> 
> I think we are definitely getting closer to release time!
> 
> The one question that might still be open is whether there is still
> work to be done regarding the "target:" default sysroot.
> 
> As far as I know, there has been some awesome work between Pedro and
> Gary to significantly reduce the time it takes to perform the file
> transfers.
> 
> There was also the question of being able to interrupt an on-going
> file tranfer, and I saw some promising patches, although Sandra reported
> that the quick-and-dirty proof-of-concept left GDB in an inconsistent
> state after interruption...

I'm having quite an interrupted day, but I'm trying to clean up
the hacky patch a bit.  Even if GDB ends up showing a couple
errors after disconnecting, like:

 You can't do that when your target is `exec'
 No registers.
 (gdb)

I think it's way better than getting stuck.  It's an escape hatch
for emergencies -- some bruising expected.  :-)  We can always clean
that up after the release, IMO.  I think those sort of errors can happen
if the connection drops unexpectedly at the wrong time, even
without the patch, for instance.

> 
> There was also a proposal to revert the default sysroot value back
> to "", with some added logic to make the "target:" implicit in some
> situations; but, as far as I know, this has been abandoned (and
> looking back, I would agree).
> 

I agree too, especially given we've fixed the major slowness
culprit.

> So, what do we still need to do before we can release?
> 
>   - Finish the work on being able to interrupt the file transfer?
> 
>   - There were some discussions about displaying a warning,
>     and I don't see anything in the current sources; are we still
>     thinking about that?

Yes, I think so.  Also Gary's QUIT patch should probably go in
as well, I think.

Thanks,
Pedro Alves

Re: GDB 7.10 release 2015-08-21 status update

[Joel Brobecker]

> I'm having quite an interrupted day, but I'm trying to clean up
> the hacky patch a bit.  Even if GDB ends up showing a couple
> errors after disconnecting, like:
> 
>  You can't do that when your target is `exec'
>  No registers.
>  (gdb)
> 
> I think it's way better than getting stuck.  It's an escape hatch
> for emergencies -- some bruising expected.  :-)  We can always clean
> that up after the release, IMO.  I think those sort of errors can happen
> if the connection drops unexpectedly at the wrong time, even
> without the patch, for instance.

Indeed, that makes sense.

> I agree too, especially given we've fixed the major slowness
> culprit.
> 
> > So, what do we still need to do before we can release?
> > 
> >   - Finish the work on being able to interrupt the file transfer?
> > 
> >   - There were some discussions about displaying a warning,
> >     and I don't see anything in the current sources; are we still
> >     thinking about that?
> 
> Yes, I think so.  Also Gary's QUIT patch should probably go in
> as well, I think.

Ah yes, I forgot about that one!

Thanks, Pedro.
-- 
Joel