[Bug 204282] [NEW PORT]: devel/gecode3: gecode version 3.7.3 is needed for compatibility for dep-selector-libgecode rubygem.

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=204282

Alexey Dokuchaev <danfe at FreeBSD.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|New                         |In Progress
                 CC|                            |danfe at FreeBSD.org
           Assignee|freebsd-ports-bugs at FreeBSD. |danfe at FreeBSD.org
                   |org                         |

--- Comment #3 from Alexey Dokuchaev <danfe at FreeBSD.org> ---
> # Created by: mokhi64 at gmail.com

While not required, it's better (more consistent with other ports) to provide
fully-qualified email address and $FreeBSD$ id tag:

> # Created by: John Doe <mokhi64 at gmail.com>
> $FreeBSD$

Portlint should catch this error by the way, but you already know this. ;-)

> COMMENT=        Generic Constraint Development Environment (Needed for dep-selector-libgecode and berkshelf)

COMMENT line is too long (again, portlint(1) should've caught that) and the the
part in parenthesis is useless, as are excessive capitalizations.  Correct
COMMENT should be something like:

> COMMENT=        Generic constraint development environment

Here, BUILD_DEPENDS typically come before LIB_DEPENDS:

> LIB_DEPENDS=    libmpfr.so:${PORTSDIR}/math/mpfr
> BUILD_DEPENDS=  bash:${PORTSDIR}/shells/bash

Also, usually build-depends on bash(1) can be easily fixed as it often due to
linuxisms, not the actual bash(1) features that to do map easily to BSD sh(1)
syntax.

> CONFIGURE_ARGS= [long list of options]

Are they really all needed?  Typically you'd only want to change the defaults,
not list them all. ;-)

> MAKE_JOBS_UNSAFE=yes

Did you try to fix it so it becomes jobs-safe?

> SHLIB=          37
> ...
> PLIST_SUB=      SHLIB="${SHLIB}"

SHLIB variable is used only once (quoted above).  Consider doing this instead:

> PLIST_SUB=      SHLIB=PORTVERSION:R

In this place in Makefile, knobs are also not well sorted (style bug).

> GIST_CONFIGURE_ON=      --enable-gist --enable-qt
> GIST_CONFIGURE_OFF=     --disable-gist --disable-qt

Consider using GIST_CONFIGURE_ENABLE knob(s) instead.

> .include <bsd.port.options.mk>
>         
> .if ${ARCH} == "amd64"
> CFLAGS+=        -D__USE_ISOC99
> .endif

Setting CFLAGS_amd64 would DTRT and you'd also won't have to include
<bsd.port.options.mk> here.

> post-install:
>         @${STRIP_CMD} ${STAGEDIR}${PREFIX}/bin/fz
> #       @${STRIP_CMD} ${STAGEDIR}${PREFIX}/lib/libgecode*.so.[0-9]*.[0-9]*

Why the last line is commented?  Also, consider using
INSTALL_TARGET=install-strip if configure script supports it.

In pkg-plist: @dirrmtry lines are not required anymore.  Now, about the patch:

> --- configure.orig
> +++ configure
> @@ -1,4 +1,4 @@
> -#! /bin/sh
> +#! /usr/bin/env bash
>  # From configure.ac Id: configure.ac.in 12614 2012-03-22 01:42:32Z tack .
>  # Guess values for system-dependent variables and create Makefiles.
>  # Generated by GNU Autoconf 2.68 for GECODE 3.7.3.

What's wrong with /bin/sh?  Why this port needs bash(1)?

-- 
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for the bug.