error in pipe implementation?

[Fabian Sturm]

Hello

i have posted some sample code about pipes a week ago.
And i also filed a bug report against menu (nothing happend so far ).

But im now sure there is some error in the pipe handling of hurd too.

As i told you did the update-menu program write to a pipe
which caused hurd to throw a SIGPIPE at the process and terminate it.

BUT the process wrote to the read file descriptor of the pipe.
And Stevenson says that SIGPIPE is only generated if i write 
to a write file descriptor of a pipe with all read file descriptors closed.
(i dont have the posix standard)

It doesnt say something about writing to the read file descriptor.

On linux we dont get a SIGPIPE (this was what caused the difference)
just the write fails with a: Bad file descriptor EBADF.

Shouldnt this also be true for hurd?

So perhaps someone could bring light into this?
Am i wrong or is there an error?

Thanks,
sincerly Fabian