Bug#197115: vrflash_0.24-3(hppa/unstable): config.guess and friends out of date

[Richard Braakman <dark@xs4all.nl>]

On Wed, Jun 11, 2003 at 11:04:02PM -0600, lamont+buildd@hp.com wrote:
> Package: vrflash
> Version: 0.24-3
> Severity: serious

(Aside: I think that use of config.sub in autoconf is nearly always a bug.
I'd rather fix the part of configure.in that thinks it needs to know the
platform name, than work around it.)
 
> > checking for a BSD compatible install... /usr/bin/install -c
> > checking host system type... Invalid configuration `parisc64-unknown-linux': machine `parisc64-unknown' not recognized

Hmm.  This check is done to set a variable called "host_x86", which is
used only once:

if test "x$have_linuxh" = "xyes" ; then
        if test "x$host_x86" = "xyes" ; then
                AC_MSG_CHECKING(for low_latency serial option)
                AC_TRY_COMPILE([#include <linux/serial.h>], \
                        [int a = ASYNC_LOW_LATENCY;], \
                        low_latency=yes,low_latency=no)
                if test "x$low_latency" = "xyes" ; then
                        AC_MSG_RESULT(yes)
                        AC_DEFINE(HAVE_LOW_LATENCY)
                else
                        AC_MSG_RESULT(no)
                fi
        fi
fi

I have no idea why it checks for x86 rather than just testing for
ASYNC_LOW_LATENCY directly.  I have only a vague idea what the
low-latency flag does (it seems to be something like "send all tty
data immediately instead of trying to batch it"), but I don't see
anything x86-specific about it.  There's nothing in the changelog
about why it tests for x86.

I don't particularly want to go fix this now, because
  a) I have no way to test the results, and
  b) I just orphaned the package for that reason.

I guess that testing it properly means finding someone with an Agenda VR
and a non-x86 machine.  In the meantime, the best approach might be to
rip out the host_x86 part of the test and see if anyone complains.
I'll leave that to the next maintainer.

You've set the severity to "serious".  Do you think anything actually
broke?  If not, I'll reduce it to "minor" (investigating a warning).
I don't know whether the use of host_x86 is correct, but an unknown
platform does correctly set that variable to "no".  This shows in the
build log, which never runs the "Checking for low latency serial option"
check.

Richard Braakman