[Mono-dev] Mono 2.6.x
gnorton at novell.com
Mon Feb 8 15:03:12 EST 2010
On 2010-02-08, at 4:57 AM, matteo tesser wrote:
> Hi Geoff,
> Is the well tested stable component you refer, public? can we help in testing?
Yes, its the GC. We've made the change in trunk which means that it will be in 2.8 barring any problems.
> I only wan to say that Its a pity that this will not back-ported.
> Given this any change of running multi-thread apps is loss.
> I noticed that (at least in my case, a 3x performances boost can be
> obtained on Mac OS X also in cases where there only one background
> working thread and an UI thread). Anyway I agree that is more
> important to have apps running slow that not running at all.
2.6 is a stable supported release, I've discussed with the runtime team, and the consensus is that backporting a fundamental change to the GC into the 2.6 branch is not worth the risks, it needs more time to incubate in trunk so that it can be included in 2.8.
> Finally a question: I noticed when running autogen.sh I saw the
> following line:
> Parallel Mark: Disabled_Currently_Hangs_On_MacOSX
> I assume that meaining of the word Hangs is: it goes too slow, and
> maybe this can solved by the synchronization speedup.
No, hangs means it hangs and deadlocks the process.
> I tried to force the use of parallel mark by compiling the trunk with
> the following option --enable-parallel-mark, and I obtain the
> following error on make, is this normal?
> Making all in mini
> make all-am
> ../../doltlibtool --quiet --tag=CC --mode=link gcc -I../..
> -I../../libgc/include -D_REENTRANT
> -I/Users/teo/mono2.5/lib/glib-2.0/include -g -O2
> -fno-strict-aliasing -g -Wall -Wunused -Wmissing-prototypes
> -Wmissing-declarations -Wstrict-prototypes -Wmissing-prototypes
> -Wnested-externs -Wpointer-arith -Wno-cast-qual -Wwrite-strings
> -export-dynamic -sectcreate __TEXT __info_plist
> ../../mono/mini/Info.plist -pthread -o mono main.o libmono-static.la
> -L/Users/teo/mono2.5/lib -lgthread-2.0 -lglib-2.0 -lintl -lm
> -lpthread -lm
> Undefined symbols:
> "_mono_gc_get_write_barrier", referenced from:
> _decode_method_ref in libmono-static.a(aot-runtime.o)
> ld: symbol(s) not found
Are you still seeing this?
> On Wed, Feb 3, 2010 at 8:29 PM, Geoff Norton <gnorton at novell.com> wrote:
>> On 2010-02-03, at 2:27 AM, matteo tesser wrote:
>>> The following bug has been fixed on the trunk, I would like it to be backported:
>> This patch will not be backported, it changes the behaviour of well tested stable component in a major way with a serious chance of regressions.
More information about the Mono-devel-list