The autotools setup detects and links against it, but the code of vim
doesn't use it. Discusse with sthen@, who found this mail (2001!) from
upstream.
https://marc.info/?l=vim-dev&m=100487203621247&w=2
On powerpc, MACHINE_ARCH is "powerpc" and ARCH is "socppc" or "macppc",
the test fails and neomutt tries to use kyotocabinet, marked as BROKEN.
Fix test and use qdbm as intented. On arm, where MACHINE_ARCH is "arm" and
ARCH is "armv7", the test fails thus neomutt is already using kyotocabinet
(which builds on arm now), so remove the test. No need for a bump since
the build failed in both cases.
Reported by "gumnos", ok sthen@ (maintainer)
- Add COMPILER and drop ONLY_FOR_ARCHS
- Add -std=c++11 (clang also needs -Wno-c++11-narrowing to build successful)
- Drop boost as bdep and change tr1-implementation to system.
- Fake function in namespace std::tr1:: and use std::placeholders.
tweak and ok bluhm@
Security patches removed, since all of them appear to be merged
upstream, APIWARN diffs kept. The atexit workaround introduced by jolan
is not needed any more.
The port now explicitely requests gnu11 support. I did not change the
feature set (no libjpeg, no opengl) to ease the update. People can
enable stuff later if they need to.
Testing quirks:
- not enabling jpeg means that the two first tests fail
- shells/bash should be installed at *configure* time in order for tests
to be picked up; maybe we should add a "test" FLAVOR so that regular
builds aren't affected?
Tested on amd64 and arm, bulk build + ok ajacoutot@ (thanks!)
based on questionable advice provided by igor(1), so clarify the
purpose and the limitations of the port in the DESCR and bump.
Using wording tweaks from tb@.
OK tb@ and no objection from jmc@.
- replace more hardcoded php binary names with ${MODPHP_BIN}, unbreaks some
scripts as we include the PHP release branch in the string (php-5.6 etc).
- drop php-mysql dependency, no longer required.
