Re: GDB 6.2 release schedule?

Re: GDB 6.2 release schedule?

[Michael Elizabeth Chastain]

mec> 1677 is probably easy to fix.  1650 looks significantly deeper.
mec> The problem with 1650 is that gdb HEAD randomly gets internal errors,
mec> but gdb 6.1 does not, on the same test script.
ac> These worry me

I'll be looking at 1650 over the weekend.  I doubt that I can fix it
but I can analyze it some more.

ac> (btw, no one got around to moving selftest.exp to 
ac> testsuite/gdb.gdb/ :-).

Is there a PR for it?

ac> But these don't. Presumably being ISA specific its possible to have 
ac> fixes easily pulled into the branch.

Good point.  Some of them are probably hppa-specific or hpux-specific
and easy to fix.

But some of the problems involve watchpoints, so it might be something
that happens on several arches besides hppa-hpux.  All I really know is
that it happens on hppa-hpux and does not happen on i686-linux.

I have to roll up my sleeves and do a bunch of cvs narrowing, which
is a pain on the hp test drive cluster.

Michael C

Re: GDB 6.2 release schedule?

[Andrew Cagney]

> mec> 1677 is probably easy to fix.  1650 looks significantly deeper.
> mec> The problem with 1650 is that gdb HEAD randomly gets internal errors,
> mec> but gdb 6.1 does not, on the same test script.
> ac> These worry me
> 
> I'll be looking at 1650 over the weekend.  I doubt that I can fix it
> but I can analyze it some more.
> 
> ac> (btw, no one got around to moving selftest.exp to 
> ac> testsuite/gdb.gdb/ :-).
> 
> Is there a PR for it?

It's just a todo item, after gdb.gdb/ was created, no one got around to 
moving selftest.exp there (the test gdb debugging gdb directory), oops :-)

> ac> But these don't. Presumably being ISA specific its possible to have 
> ac> fixes easily pulled into the branch.
> 
> Good point.  Some of them are probably hppa-specific or hpux-specific
> and easy to fix.
> 
> But some of the problems involve watchpoints, so it might be something
> that happens on several arches besides hppa-hpux.  All I really know is
> that it happens on hppa-hpux and does not happen on i686-linux.

FYI,JeffJ's already narrowed the watchpoint regressions down to a 
specific patch, so I think that's under control.

> I have to roll up my sleeves and do a bunch of cvs narrowing, which
> is a pain on the hp test drive cluster.

Andrew

Re: GDB 6.2 release schedule?

[Michael Elizabeth Chastain]

I think July 1 is too soon to cut the branch.

native i686-pc-linux-gnu has two regressions from gdb 6.1 to gdb HEAD.

  gdb/1650  manythreads.exp
  gdb/1677  [regression] selftest.exp, crash in lookup_symtab

1677 is probably easy to fix.  1650 looks significantly deeper.
The problem with 1650 is that gdb HEAD randomly gets internal errors,
but gdb 6.1 does not, on the same test script.

native hppa2.0w-hp-hpux11.11 has regressions in 17 test scripts.
I estimate there's about 1 real regression bug per 4-5 test scripts,
so there are about 4-5 real regression bugs to hunt down and fix
(or document).

  http://sources.redhat.com/ml/gdb-testers/2004-q2/msg00160.html
  http://sources.redhat.com/ml/gdb-testers/2004-q2/msg00157.html

Michael C

Re: GDB 6.2 release schedule?

[Andrew Cagney]

> I think July 1 is too soon to cut the branch.
> 
> native i686-pc-linux-gnu has two regressions from gdb 6.1 to gdb HEAD.
> 
>   gdb/1650  manythreads.exp
>   gdb/1677  [regression] selftest.exp, crash in lookup_symtab
> 
> 1677 is probably easy to fix.  1650 looks significantly deeper.
> The problem with 1650 is that gdb HEAD randomly gets internal errors,
> but gdb 6.1 does not, on the same test script.

These worry me (btw, no one got around to moving selftest.exp to 
testsuite/gdb.gdb/ :-).

> native hppa2.0w-hp-hpux11.11 has regressions in 17 test scripts.
> I estimate there's about 1 real regression bug per 4-5 test scripts,
> so there are about 4-5 real regression bugs to hunt down and fix
> (or document).
> 
>   http://sources.redhat.com/ml/gdb-testers/2004-q2/msg00160.html
>   http://sources.redhat.com/ml/gdb-testers/2004-q2/msg00157.html

But these don't. Presumably being ISA specific its possible to have 
fixes easily pulled into the branch.

There's also a balancing act here.  GDB as a whole should continue to 
improve.  While there are the above regressioins, there are also 
dramatic improvements on other ISAs such as i386 and PPC/PPC64.

enjoy,
Andrew

Re: GDB 6.2 release schedule?

[Randolph Chung]

> native hppa2.0w-hp-hpux11.11 has regressions in 17 test scripts.
> I estimate there's about 1 real regression bug per 4-5 test scripts,
> so there are about 4-5 real regression bugs to hunt down and fix
> (or document).

Looks like Jeff's watchpoint fix fixes the recurse.exp regression
against 6.1(.1) for hppa-hpux. it probably will also fix some of the
other regressions...  very cool :)

recurse.exp is still broken on hppa-linux :( It works with stabs but not
with dwarf2.  Of course, that is not a regression....

i can look at some of the other regressions next week.

randolph
-- 
Randolph Chung
Debian GNU/Linux Developer, hppa/ia64 ports
http://www.tausq.org/

GDB 6.2 release schedule?

[Andrew Cagney]

I've come up the following dates:
   Branch:  July 1 (2004-07)
   Release: August (2004-08)
   reSpin:  September (2004-09)
http://www.gnu.org/software/gdb/schedule/

There's clearly a desire to at least have a new GDB containing all the 
recent PPC changes come out sooner rather than later.

So please discuss :-)

Andrew