Re: Package building on xen- or kvm-based buildds?
On Thu, 2011-12-01 at 18:40 +0100, Samuel Thibault wrote:
> Svante Signell, le Thu 01 Dec 2011 17:54:40 +0100, a écrit :
> > On Thu, 2011-12-01 at 17:32 +0100, Richard Braun wrote:
> > > On Thu, Dec 01, 2011 at 05:14:09PM +0100, Svante Signell wrote:
> > > > According to Pino T. a sysv-package can be built when shared memory is
> > > > implemented (I thought it was already available?)
> > >
> > > Don't mistake "shared memory" with "System V shared memory" (sysvshm).
> > > It's about the interface, not the functionality.
> > So the SysV functions are not yet implemented, like shmget, shmat,
> > shmctl, etc.
> These *are* implemented, that's why the difference with SysV IPC and
> semaphores is important :)
Sorry that I don't get it: Pino said on November 30:
22:17:19< gnu_srs1> How much work is needed to support fakeroot-sysv?
22:17:50< pinotree> implement sysv shared memory
And you claim that it is implemented already, so why do we use
fakeroot-tcp, and not fakeroot-sysv?
>> and fakeroot-sysv use SysV
>> shared memory, right?
>No, it uses IPC and semaphores.
>> Another puzzling thing:
>> Since semget, etc is not yet implemented, which semaphores are
>> implemented in Hurd then?
>Only in-process semaphores, by libpthread. I.e. sem_init, but not
Q1) What is the relation between fakeroot-sysv and SysV shared memory?
Q2) fakeroot-sysv use IPC and semaphores. IPC is supported, but only
for in-process semaphores are: Complete semaphore support is needed for
fakeroot-sysv to work properly?
Q3) What is the current status of fakeroot-hurd? Would that be a better
alternative for the wholee fakeroot issue?