[Mesa-dev] [PATCH 1/4] configure.ac: Define HAVE_ZLIB if zlib is found
Jordan Justen
jordan.l.justen at intel.com
Fri Nov 10 20:35:35 UTC 2017
On 2017-11-10 11:38:08, Emil Velikov wrote:
> On 10 November 2017 at 19:34, Emil Velikov <emil.l.velikov at gmail.com> wrote:
> > On 10 November 2017 at 18:38, Dylan Baker <dylan at pnwbakers.com> wrote:
> >> Quoting Jordan Justen (2017-11-10 10:24:36)
> >>> Cc: Emil Velikov <emil.velikov at collabora.com>
> >>> Cc: Dylan Baker <dylan at pnwbakers.com>
> >>> Signed-off-by: Jordan Justen <jordan.l.justen at intel.com>
> >>> ---
> >>> configure.ac | 6 +++++-
> >>> meson.build | 6 +++++-
> >>> 2 files changed, 10 insertions(+), 2 deletions(-)
> >>>
> >>> diff --git a/configure.ac b/configure.ac
> >>> index 411c4f6b3e0..213cfe132fb 100644
> >>> --- a/configure.ac
> >>> +++ b/configure.ac
> >>> @@ -900,7 +900,11 @@ dnl See if posix_memalign is available
> >>> AC_CHECK_FUNC([posix_memalign], [DEFINES="$DEFINES -DHAVE_POSIX_MEMALIGN"])
> >>>
> >>> dnl Check for zlib
> >>> -PKG_CHECK_MODULES([ZLIB], [zlib >= $ZLIB_REQUIRED])
> >>> +PKG_CHECK_MODULES([ZLIB], [zlib >= $ZLIB_REQUIRED],
> >>> + [have_zlib=yes], [have_zlib=no])
> >>> +if test "x$have_zlib" = xyes; then
> >>> + DEFINES="$DEFINES -DHAVE_ZLIB"
> >>> +fi
> >>>
> >>> dnl Check for pthreads
> >>> AX_PTHREAD
> >>> diff --git a/meson.build b/meson.build
> >>> index 855c80ed658..4c27b3ba048 100644
> >>> --- a/meson.build
> >>> +++ b/meson.build
> >>> @@ -661,8 +661,12 @@ if dep_libdrm.found()
> >>> endif
> >>> endif
> >>>
> >>> +dep_zlib = dependency('zlib', version : '>= 1.2.3', required : false)
> >>> +if dep_zlib.found()
> >>> + pre_args += '-DHAVE_ZLIB'
> >>> +endif
> >>> +
> >>
> >> There's the gen_zipped_file.py in src/intel/genxml that uses zlib for something,
> >> is not having zlib going to break that?
> >>
> > Those are the aubinator* family of tools. But more importantly, we
> > have compression of the disk cache.
> >
> > One solution is to opt for the disk cache approach - ifdef the code
> > and build it only on !Windows platforms.
> > There should be others, but ^^ sounds like the shortest one.
> >
> Having read through the rest of the series:
>
> The current HAVE_ZLIB approach looks good, modulo:
> - keep zlib a hard requirement in autotools/meson, explicitly setting
> the define
Ok. I'll make this change.
Eventually, I think meson may replace scons, but it sounds like meson
has an easy way to pull in zlib on windows. I'll leave that to Dylan.
:)
> - add missing guards in src/intel - only if you want to since the
> code is build on !Windows platforms anyway
I think I'll skip this for now. It'd need to be sent into the python
script, and really doesn't seem worth it. :)
-Jordan
More information about the mesa-dev
mailing list