On Sat, Jan 30, 2010 at 7:28 PM, Lars Luthman
<address@hidden> wrote:
On Thu, 2010-01-28 at 14:14 +0200, Marius Stoica wrote:
> Hi,
>
> I managed to complie dino on puppy linux 431 with jack 0.118.0 and
> lash 0.5.4 (using the 0.2.2 tarball patched for the 27510 bug)
> dino starts ok, lash works fine also, but jack transport is stuck
> (because of dino).
> I can see in qjackctl that the bpm is 0 and any change made in dino
> for the bpm wil give an abnormal
> number in qjackctl and the transport remains stuck.
> I believe this has something to do with the jack transport code in
> dino, maybe it must be updated to the latest jack api.
I think I know what the problem might be now. It's an old bug that was
fixed in the Git repository in april last year, I had forgotten that it
was still in the latest release.
Technical description: JACK uses a struct type called jack_position_t to
store the current position and tempo, and Dino accessed some members in
it by reference. At some point the JACK hackers changed this struct to
be declared with __attribute__((__packed__)), probably to make the
struct layout the same for 32-bit and 64-bit clients on a mixed system.
This changed the byte offsets of the struct members so that the
references that Dino used are no longer valid. This _should_ have been
caught by the compiler, but apparently it wasn't.
Short description: The JACK ABI changed. Dino broke.
Still, this has been fixed in the Git repository for a while, so the
current head of the branch Branch_0_2 should work (unless there are
other bugs that I haven't seen). You can get this either by running the
commands
git checkout -b Branch_0_2 origin/Branch_0_2
git pull
in your clone of the Dino Git repository, or by downloading a snapshot
from
http://git.savannah.gnu.org/cgit/dino.git/snapshot/dino-Branch_0_2.tar.gz
You could also wait a couple of minutes while I test and release 0.2.3.
--ll