Comment 48 for bug 219630

I'm not happy to see the debian build system changes in the -nsc SRU; they should be almost entirely irrelevant, but in the process they still introduce a risk of regression. This is not something that should ever be included in an SRU, and if I approve it it's only because I feel I'm wasting my time trying to get a minimal diff when this point has already been made to no avail, and we might as well accept the package into -proposed and regression-test it there.

Given the late landing, therefore, I believe this SRU should be held until after 8.04.1 is buildable from hardy-updates only.

BTW, it's also very confusing to have a patch which changes, and then /comments out/ a line; it took me far too long to parse this:

-+ awk '/^#define.*PCI_CHIP/ {print $$3}' ${srcdir}/nsc_driver.c | sed -e 's/0x/1078/' > nsc.ids
-+ awk '/^#define.*PCI_CHIP/ {print $$3}' ${srcdir}/nsc_driver.c | sed -e 's/0x/100B/' >> nsc.ids
++ awk '/^#define.*PCI_CHIP/ {print $$3}' ${srcdir}/nsc_driver.c | sort -u | grep -v 0030 | sed -e 's/0x/100B/' > nsc.ids
++# awk '/^#define.*PCI_CHIP/ {print $$3}' ${srcdir}/nsc_driver.c | sort -u | grep -v 0030 | sed -e 's/0x/1078/' >> nsc.ids

because I kept overlooking the '#' at the beginning of that line...

Since this changes which driver is being used for a set of chips, which AIUI aren't unsupported by nsc they're just supported "better" by the cyrix driver, SRU validation needs to include feedback from users of these precise chips. (Ideally, the SRU justification would also include explicit pointers to the problems with using the nsc driver instead of the cyrix driver, which AFAICS is completely absent here.) This also contributes to making this an inappropriate change to include at the last minute before 8.04.1.