Hello there! Apparently DConf had a good effect for sociomantic, as we are starting to analyze moving forward to D2. The first step would be to port CDGC to druntime (we will post some more details about this next week).

I have a couple of questions before we start.

1) I saw there is a 'benchmark' directory in druntime now. Is that
   maintained and meaningful. Should I use it?
2) Are there any tests focusing on the GC I can rely on?
3) Is each commit in druntime being tested by the autotester, or only
   DMD is using the autotester for now?
4) I would like to do also some code reorganization and cleanup in the
   GC to make it look more like CDGC[1]. Will you be opened to that or
   should I only focus on the particular changes to make the GC
   concurrent?

Any other suggestions or rules on hacking druntime are appreciated, since last time I touched it was when the project was created :)

Thanks!

[1] http://www.dsource.org/projects/tango/browser/trunk/tango/core/rt/gc/cdgc

-- 
Leandro Lucarella
Senior R&D Developer, Sociomantic Labs GmbH, http://www.sociomantic.com
_______________________________________________
D-runtime mailing list
D-runtime@puremagic.com
http://lists.puremagic.com/mailman/listinfo/d-runtime

This is great news!

Regarding refactoring - it would be great if it were a separate step, either before or after the material changes.

Andrei

On 5/29/13 12:20 PM, Leandro Lucarella wrote:
> Hello there! Apparently DConf had a good effect for sociomantic, as we
> are starting to analyze moving forward to D2. The first step would be to
> port CDGC to druntime (we will post some more details about this next
> week).
>
> I have a couple of questions before we start.
>
> 1) I saw there is a 'benchmark' directory in druntime now. Is that
>     maintained and meaningful. Should I use it?
> 2) Are there any tests focusing on the GC I can rely on?
> 3) Is each commit in druntime being tested by the autotester, or only
>     DMD is using the autotester for now?
> 4) I would like to do also some code reorganization and cleanup in the
>     GC to make it look more like CDGC[1]. Will you be opened to that or
>     should I only focus on the particular changes to make the GC
>     concurrent?
>
> Any other suggestions or rules on hacking druntime are appreciated,
> since last time I touched it was when the project was created :)
>
> Thanks!
>
> [1] http://www.dsource.org/projects/tango/browser/trunk/tango/core/rt/gc/cdgc
>
_______________________________________________
D-runtime mailing list
D-runtime@puremagic.com
http://lists.puremagic.com/mailman/listinfo/d-runtime

On Wednesday, May 29, 2013 18:20:43 Leandro Lucarella wrote:
> 3) Is each commit in druntime being tested by the autotester, or only DMD is using the autotester for now?

The autotester builds and runs the tests for dmd, druntime, and Phobos after every commit, and the pull tester does the same for every pull request which was submitted by someone on the whitelist.

- Jonathan M Davis
_______________________________________________
D-runtime mailing list
D-runtime@puremagic.com
http://lists.puremagic.com/mailman/listinfo/d-runtime

On Wed, May 29, 2013 at 01:25:52PM -0400, Andrei Alexandrescu wrote:
> This is great news!
> 
> Regarding refactoring - it would be great if it were a separate step, either before or after the material changes.

Yes, the idea is not to send a huge commit with everything but to send small commits that are easy to review (test, and debug ;).

-- 
Leandro Lucarella
Senior R&D Developer
Sociomantic Labs GmbH <http://www.sociomantic.com>
_______________________________________________
D-runtime mailing list
D-runtime@puremagic.com
http://lists.puremagic.com/mailman/listinfo/d-runtime

On Wed, May 29, 2013 at 02:53:31PM -0400, Jonathan M Davis wrote:
> On Wednesday, May 29, 2013 18:20:43 Leandro Lucarella wrote:
> > 3) Is each commit in druntime being tested by the autotester, or only DMD is using the autotester for now?
> 
> The autotester builds and runs the tests for dmd, druntime, and Phobos after every commit, and the pull tester does the same for every pull request which was submitted by someone on the whitelist.

Cool! Any other comments about 1) and 2). From Andrei's reply I assume
4) is OK.

-- 
Leandro Lucarella
Senior R&D Developer
Sociomantic Labs GmbH <http://www.sociomantic.com>
_______________________________________________
D-runtime mailing list
D-runtime@puremagic.com
http://lists.puremagic.com/mailman/listinfo/d-runtime

On 29.05.2013 18:20, Leandro Lucarella wrote:
> Hello there! Apparently DConf had a good effect for sociomantic, as we
> are starting to analyze moving forward to D2. The first step would be to
> port CDGC to druntime (we will post some more details about this next
> week).

Very much looking forward to this :-)

>
> I have a couple of questions before we start.
>
> 1) I saw there is a 'benchmark' directory in druntime now. Is that
>     maintained and meaningful. Should I use it?

IMO the files in that directory are mere tests for some functionality and don't really qualify as benchmarks. When I did tests some time ago I added a few more files to the directory:

- http://forum.dlang.org/thread/jc2cbj$1l90$1@digitalmars.com
- http://stackoverflow.com/questions/8782627/does-the-d-garbage-collector-work
- testgc2.d and testgc3.d from the test-suite

My main benchmark for performance and correctness was the parser from Visual D though.


_______________________________________________
D-runtime mailing list
D-runtime@puremagic.com
http://lists.puremagic.com/mailman/listinfo/d-runtime

On Thu, May 30, 2013 at 12:28:11PM +0200, Rainer Schuetze wrote:
> On 29.05.2013 18:20, Leandro Lucarella wrote:
> >Hello there! Apparently DConf had a good effect for sociomantic, as we are starting to analyze moving forward to D2. The first step would be to port CDGC to druntime (we will post some more details about this next week).
> 
> Very much looking forward to this :-)
> 
> >
> >I have a couple of questions before we start.
> >
> >1) I saw there is a 'benchmark' directory in druntime now. Is that
> >    maintained and meaningful. Should I use it?
> 
> IMO the files in that directory are mere tests for some functionality and don't really qualify as benchmarks. When I did tests some time ago I added a few more files to the directory:
> 
> - http://forum.dlang.org/thread/jc2cbj$1l90$1@digitalmars.com
> - http://stackoverflow.com/questions/8782627/does-the-d-garbage-collector-work
> - testgc2.d and testgc3.d from the test-suite
> 
> My main benchmark for performance and correctness was the parser from Visual D though.

If you could keep an eye on my patches with that it will be extremely appreciated, since our applications won't be migrated until this is working, I don't have any real code to test the performance of my changes.

-- 
Leandro Lucarella
Senior R&D Developer
Sociomantic Labs GmbH <http://www.sociomantic.com>
_______________________________________________
D-runtime mailing list
D-runtime@puremagic.com
http://lists.puremagic.com/mailman/listinfo/d-runtime