October 15, 2013
On Tue, Oct 8, 2013 at 11:00 PM, David Nadlinger <code@klickverbot.at> wrote:
> As discussed, here are the first alpha packages for the 0.12.0 release.

Okay, all the issues/regressions with a test case have been fixed in Git master, so I started preparations for the first beta release (#513 didn't make it in time, but hopefully we'll be able to tackle it before the final release).

I had to make some rather big changes to the internals in order to fix some of the bugs, so we unfortunately have to expect new issues. Still, I hope that we can push out the release this weekend.

Stay tuned for the announcement, including source/binary packages and the updated changelog, in a couple of hours.

Cheers,
David
October 17, 2013
On Wednesday, 9 October 2013 at 04:58:20 UTC, Kai Nacke wrote:
> On Wednesday, 9 October 2013 at 03:15:41 UTC, David Nadlinger wrote:
>> On Wed, Oct 9, 2013 at 4:55 AM, Elvis Zhou <elvis.x.zhou@gmail.com> wrote:
>>> Nice work!
>>> I'm very interested in "LDC builds druntime, Phobos and a Hello World
>>> program on Linux/ARM", is there a guidance? I wanna give it a try and see if
>>> I can be of any help to ARM support.
>>
>> What platform are you on?
>
> I still did not test the latest changes but I have a ARM-based NAS running Gentoo Linux. LDC can be installed as any other Linux program on this box.
>
> Kai

To improve development I ordered an IFC6410 (http://www.inforcecomputing.com/product/moreinfo/ifc6410.html): 4 ARM cores @1.7 GHz, 2GB RAM and SATA port. I hope I can really develop with this board (my NAS is really slow....)

Kai
October 17, 2013
On Tue, Oct 8, 2013 at 11:35 PM, bearophile <bearophileHUGS@lycos.com> wrote:
> This little program still shows a significant loss of performance compared
> to dmd:
> http://rosettacode.org/wiki/Zebra_puzzle#Alternative_Version

Sorry, I think I never replied to this so far: I had a brief look at this, and it seems like this isn't really a single codegen bug, but rather LLVM being too smart for its own good and rearranging things so there is an awful lot of branch mispredictions in the inner loop.

It would be nice if somebody else could have a look at this, though, as my copy of VTune is acting up…

David
October 17, 2013
On 17 Oct 2013, at 15:15, Kai Nacke wrote:
> To improve development I ordered an IFC6410 (http://www.inforcecomputing.com/product/moreinfo/ifc6410.html): 4 ARM cores @1.7 GHz, 2GB RAM and SATA port. I hope I can really develop with this board (my NAS is really slow....)

That seems like a nice system to work on indeed, and quite a bit more powerful than the ODROID-X I use (4 x 1.4 GHz Cortex-A9, 1 GB RAM). I think the SATA port will be particularly nice, as I/O performance tends to be the bottleneck on my board. But even given that, I tend to just compile everything on there, since it is fast enough that setting up a cross-compilation environment wasn't really worth the hassle so far.

Flash wear is really noticeable when using SD cards or eMMC modules for development, so that's another plus point. Unfortunately, there weren't really any affordable boards with (e)SATA available at the time I bought mine.

David
1 2 3 4
Next ›   Last »