GDB 8.3 release -- 2019-01-11 update

GDB 8.3 release -- 2019-01-11 update

[Joel Brobecker]

Hello again,

New update for the GDB-8.3 release. Reminder: The target date
for creating the branch is:

        Tue Jan 15th

The wiki page for tracking known issues is at:

    https://sourceware.org/gdb/wiki/GDB_8.3_Release

Fixed since the last update:

  * [JoelB] <NO PR>
    GDB crash re-running program on Windows

        Fix in review:
        https://www.sourceware.org/ml/gdb-patches/2018-12/msg00443.html

  * [KeithS] gdb/23712 (**inherited from 8.2.1**)
    dw2_add_symbol_to_list language assertion

    https://www.sourceware.org/ml/gdb-patches/2019-01/msg00231.html
    Well done, Keith, and thanks to everyone who participated in
    all the discussions!

  * [SimonM] gdb/23893
    Search paths for debug files for remote debuggee checks are broken
    https://sourceware.org/bugzilla/show_bug.cgi?id=23893

Known blocking issues:

  * [TomT/Joel] <NO PR>
    GDB UI word wrapping broken/interferes with 'set style'
    https://www.sourceware.org/ml/gdb-patches/2018-12/msg00435.html
    https://www.sourceware.org/ml/gdb-patches/2019-01/msg00098.html

    From what I can tell, "set style enabled off" doesn't prevent
    the problem from happening either. Tom indicated in his last
    message he has a fix, and just needs a testcase.

    I will try to run it against AdaCore's testsuite, which hit
    that issue quite a bit, so probably a good testbed too.

Looking at the PRs marked 8.3 as the milestone, I also see...

  * [TomT] tui/20819
    Binding Control-j in inputrc breaks "layout asm"
    https://sourceware.org/bugzilla/show_bug.cgi?id=20819

    ... but i don't understand why this would be blocking for 8.3.
    It would be nice to have a way to get a more detailed history
    of the changes made to the status of that PR so we can determine
    when it was marked 8.3. My suspicion is that it was marked 8.3
    when fixed by Tom, and then the PR got re-opened (around Jan 2nd)
    due to a regression.

    Tom - do you confirm this is not a critical bug, other than
    maybe the regression, in which case it would be sufficient to
    simply revert?

Any other issue, please let us know...

Thank you!
-- 
Joel

Re: GDB 8.3 release -- 2019-01-11 update

[Eli Zaretskii]

> Date: Fri, 11 Jan 2019 13:53:32 +0400
> From: Joel Brobecker <brobecker@adacore.com>
> 
> New update for the GDB-8.3 release. Reminder: The target date
> for creating the branch is:
> 
>         Tue Jan 15th
> 
> The wiki page for tracking known issues is at:
> 
>     https://sourceware.org/gdb/wiki/GDB_8.3_Release

Will this be from master or from the same branch where 8.2 and 8.2.1
were produced?

(I thought the next release will be 9.1, not 8.3, that's why I ask.)

Re: GDB 8.3 release -- 2019-01-11 update

[Joel Brobecker]

> > New update for the GDB-8.3 release. Reminder: The target date
> > for creating the branch is:
> > 
> >         Tue Jan 15th
> > 
> > The wiki page for tracking known issues is at:
> > 
> >     https://sourceware.org/gdb/wiki/GDB_8.3_Release
> 
> Will this be from master or from the same branch where 8.2 and 8.2.1
> were produced?

It will be from master (and thus will include the change you made
for stepping out of main on Windows, for instance).

> (I thought the next release will be 9.1, not 8.3, that's why I ask.)

That will be next release branch after 8.3 (i.e the decision was
to do one last release with the current version number, for several
practical reasons), and then switch over to the new one this summer.

-- 
Joel

Re: GDB 8.3 release -- 2019-01-11 update

[Eli Zaretskii]

> Date: Fri, 11 Jan 2019 17:40:55 +0400
> From: Joel Brobecker <brobecker@adacore.com>
> Cc: gdb-patches@sourceware.org, tom@tromey.com
> 
> > >     https://sourceware.org/gdb/wiki/GDB_8.3_Release
> > 
> > Will this be from master or from the same branch where 8.2 and 8.2.1
> > were produced?
> 
> It will be from master (and thus will include the change you made
> for stepping out of main on Windows, for instance).

OK, thanks, for clearing out my confusion.

Re: GDB 8.3 release -- 2019-01-11 update

[Philippe Waroquiers]

On Fri, 2019-01-11 at 13:53 +0400, Joel Brobecker wrote:
> Hello again,
> 
> New update for the GDB-8.3 release. Reminder: The target date
> for creating the branch is:
> 
>         Tue Jan 15th
> 
> The wiki page for tracking known issues is at:
> 
>     https://sourceware.org/gdb/wiki/GDB_8.3_Release
> 
> Fixed since the last update:
> 
>   * [JoelB] <NO PR>
>     GDB crash re-running program on Windows
> 
>         Fix in review:
>         https://www.sourceware.org/ml/gdb-patches/2018-12/msg00443.html
> 
>   * [KeithS] gdb/23712 (**inherited from 8.2.1**)
>     dw2_add_symbol_to_list language assertion
> 
>     https://www.sourceware.org/ml/gdb-patches/2019-01/msg00231.html
>     Well done, Keith, and thanks to everyone who participated in
>     all the discussions!
> 
>   * [SimonM] gdb/23893
>     Search paths for debug files for remote debuggee checks are broken
>     https://sourceware.org/bugzilla/show_bug.cgi?id=23893
> 
> Known blocking issues:
> 
>   * [TomT/Joel] <NO PR>
>     GDB UI word wrapping broken/interferes with 'set style'
>     https://www.sourceware.org/ml/gdb-patches/2018-12/msg00435.html
>     https://www.sourceware.org/ml/gdb-patches/2019-01/msg00098.html
> 
>     From what I can tell, "set style enabled off" doesn't prevent
>     the problem from happening either. Tom indicated in his last
>     message he has a fix, and just needs a testcase.
I did some manual test in a corner with the patch.  The patch
improves significantly, but there are still one or two things
 not ok (see msg00098.html thread).

Regarding the style, it would be really nice to have:
https://sourceware.org/ml/gdb-patches/2018-12/msg00449.html
as otherwise, set|show style online help/behaviour is
inconsistent with the other set|show.

It would be nice to have
https://sourceware.org/ml/gdb-patches/2019-01/msg00227.html
(as GDB will color more file names, functions
and addresses).

Thanks
Philippe


> 
>     I will try to run it against AdaCore's testsuite, which hit
>     that issue quite a bit, so probably a good testbed too.
> 
> Looking at the PRs marked 8.3 as the milestone, I also see...
> 
>   * [TomT] tui/20819
>     Binding Control-j in inputrc breaks "layout asm"
>     https://sourceware.org/bugzilla/show_bug.cgi?id=20819
> 
>     ... but i don't understand why this would be blocking for 8.3.
>     It would be nice to have a way to get a more detailed history
>     of the changes made to the status of that PR so we can determine
>     when it was marked 8.3. My suspicion is that it was marked 8.3
>     when fixed by Tom, and then the PR got re-opened (around Jan 2nd)
>     due to a regression.
> 
>     Tom - do you confirm this is not a critical bug, other than
>     maybe the regression, in which case it would be sufficient to
>     simply revert?
> 
> Any other issue, please let us know...
> 
> Thank you!

Re: GDB 8.3 release -- 2019-01-11 update

[Tom Tromey]

>>>>> "Joel" == Joel Brobecker <brobecker@adacore.com> writes:

Joel>   * [TomT/Joel] <NO PR>
Joel>     GDB UI word wrapping broken/interferes with 'set style'
Joel>     https://www.sourceware.org/ml/gdb-patches/2018-12/msg00435.html
Joel>     https://www.sourceware.org/ml/gdb-patches/2019-01/msg00098.html

I have a patch for this, I just need to write a test.
Also I think I know what is up with the subsequent problems that
Philippe reported.

Joel>   * [TomT] tui/20819
Joel>     Binding Control-j in inputrc breaks "layout asm"
Joel>     https://sourceware.org/bugzilla/show_bug.cgi?id=20819
Joel>     ... but i don't understand why this would be blocking for 8.3.
Joel>     It would be nice to have a way to get a more detailed history
Joel>     of the changes made to the status of that PR so we can determine
Joel>     when it was marked 8.3. My suspicion is that it was marked 8.3
Joel>     when fixed by Tom, and then the PR got re-opened (around Jan 2nd)
Joel>     due to a regression.

Joel>     Tom - do you confirm this is not a critical bug, other than
Joel>     maybe the regression, in which case it would be sufficient to
Joel>     simply revert?

Reverting is certainly possible but perhaps it can be fixed some other
way.  The problem is that in "nonl" mode, RET in readline won't print a
newline, leading to gdb output being on the same line as the input.  I
was thinking maybe the RET key binding, or maybe some other function in
TUI, could print a newline when the input is accepted.

Tom

Re: GDB 8.3 release -- 2019-01-11 update

[Joel Brobecker]

> Joel>   * [TomT/Joel] <NO PR>
> Joel>     GDB UI word wrapping broken/interferes with 'set style'
> Joel>     https://www.sourceware.org/ml/gdb-patches/2018-12/msg00435.html
> Joel>     https://www.sourceware.org/ml/gdb-patches/2019-01/msg00098.html
> 
> I have a patch for this, I just need to write a test.
> Also I think I know what is up with the subsequent problems that
> Philippe reported.

Great! When you have something you'd like me to test, I'm quite happpy
to give it a try and report back.

> Joel>   * [TomT] tui/20819
> Joel>     Binding Control-j in inputrc breaks "layout asm"
> Joel>     https://sourceware.org/bugzilla/show_bug.cgi?id=20819
> Joel>     ... but i don't understand why this would be blocking for 8.3.
> Joel>     It would be nice to have a way to get a more detailed history
> Joel>     of the changes made to the status of that PR so we can determine
> Joel>     when it was marked 8.3. My suspicion is that it was marked 8.3
> Joel>     when fixed by Tom, and then the PR got re-opened (around Jan 2nd)
> Joel>     due to a regression.
> 
> Joel>     Tom - do you confirm this is not a critical bug, other than
> Joel>     maybe the regression, in which case it would be sufficient to
> Joel>     simply revert?
> 
> Reverting is certainly possible but perhaps it can be fixed some other
> way.  The problem is that in "nonl" mode, RET in readline won't print a
> newline, leading to gdb output being on the same line as the input.  I
> was thinking maybe the RET key binding, or maybe some other function in
> TUI, could print a newline when the input is accepted.

I see you just sent a patch. This is even better :). I suggested
the reverting only because you mentioned it in the PR.

-- 
Joel

Re: GDB 8.3 release -- 2019-01-11 update

[Joel Brobecker]

> >   * [TomT/Joel] <NO PR>
> >     GDB UI word wrapping broken/interferes with 'set style'
> >     https://www.sourceware.org/ml/gdb-patches/2018-12/msg00435.html
> >     https://www.sourceware.org/ml/gdb-patches/2019-01/msg00098.html
> > 
> >     From what I can tell, "set style enabled off" doesn't prevent
> >     the problem from happening either. Tom indicated in his last
> >     message he has a fix, and just needs a testcase.
> I did some manual test in a corner with the patch.  The patch
> improves significantly, but there are still one or two things
>  not ok (see msg00098.html thread).

Thanks for doing that, Philippe.

> Regarding the style, it would be really nice to have:
> https://sourceware.org/ml/gdb-patches/2018-12/msg00449.html
> as otherwise, set|show style online help/behaviour is
> inconsistent with the other set|show.

FTR (helps me when I do these status updates), this is now checked in:
https://www.sourceware.org/ml/gdb-patches/2019-01/msg00296.html

> It would be nice to have
> https://sourceware.org/ml/gdb-patches/2019-01/msg00227.html
> (as GDB will color more file names, functions
> and addresses).

Agreed; I would not consider these a blocker for creating the branch,
however. Normally, these are normally easily backported.

Thanks for the feedback!
-- 
Joel