[arch-general] Patching in PKGBUILDS

[Jan Alexander Steffens]

On Wed, Jan 1, 2014 at 5:17 AM, Mark Lee <mark at markelee.com> wrote:
> On Tue, 2013-12-31 at 23:02 -0500, Mark Lee wrote:
>> On Wed, 2014-01-01 at 04:58 +0100, Jan Alexander Steffens wrote:
>> > On Wed, Jan 1, 2014 at 4:51 AM, Mark Lee <mark at markelee.com> wrote:
>> > > My patch command was given in the build() section of the PKGBUILD:
>> > >
>> > > build() {
>> > >   cd "${srcdir}/${pkgname}-${pkgver}";
>> > >   patch -Np1 -i "${srcdir}/i965_rendering.patch"
>> > >   ./configure --prefix=/usr
>> > >   make
>> > > }
>> > >
>> > > While debugging libva-intel-driver, I applied the rendering patch via
>> > > the PKGBUILD only to find later that it wasn't actually applied. Makepkg
>> > > didn't stop building the package even though it failed to apply the
>> > > patch.
>> >
>> > Nothing wrong with that snippet; the problem lies somewhere else. Did
>> > you actually watch the patch fail and makepkg continue? Are you sure
>> > the package you have installed came from this PKGBUILD?
>>
>> I rebuilt and installed the package from the PKGBUILD (makepkg -s -r -i)
>> multiple times. After finding that nothing was working, I grew
>> suspicious and examined the src folder of the PKGBUILD to find that the
>> source was not patched. Incidentally the correct code was "patch -p1 -i"
>> not "patch -Np1 -i".
>>
>> Regards,
>> Mark
>
> Salutations,
>
> Scratch that, I just checked my records and I used a lower cased "n"
> instead of "N". I just tried it with "patch -Np1 -i" and it works, but
> "patch -np1 -i" doesn't.

Yeah, with that option patch would interpret a unified diff as a no-op.