[MacPorts] #43579: zathura-plugin-pdf-poppler @0.2.5

[MacPorts]

#43579: zathura-plugin-pdf-poppler @0.2.5
-----------------------------------------+--------------------------------
  Reporter:  starkhalo@…                 |      Owner:  macports-tickets@…
      Type:  submission                  |     Status:  new
  Priority:  Normal                      |  Milestone:
 Component:  ports                       |    Version:  2.2.1
Resolution:                              |   Keywords:
      Port:  zathura-plugin-pdf-poppler  |
-----------------------------------------+--------------------------------

Comment (by starkhalo@…):

 Replying to [comment:2 ryandesign@…]:
 > Thanks. Some observations about the Portfile:
 >
 >  * There should be a blank line between the "`# $Id$`" line and the
 "`PortSystem 1.0`" line; running "`port lint --nitpick`" should mention
 that.

 Yup, fixed the warnings. `port lint` is mentioned at §3.1.18 ''lint''
 https://guide.macports.org/index.html#using.port.lint, I missed it.
 Perhaps it should be reminded of at §4.7 ''Portfile Best Practices''?

 >  * The license field should be just "`zlib`", not "`zlib License`" (the
 license field is a space-separated list of values; consult the
 port_binary_distributable.tcl script for a non-exhaustive list of license
 names)

 Fixed, the guide at §5.1 ''Global keywords''
 https://guide.macports.org/index.html#reference.keywords makes no mention
 of this, should I open a ticket?

 >  * Checksums are not used for patchfiles that will be copied to the
 files directory; they are used for patchfiles that get downloaded, but we
 only do that in the very unusual circumstance that a patchfile is provided
 by a third party and it is too large to be conveniently copied into the
 files directory.

 Removed. `port distfiles PORTNAME` gives reason to think patchfiles
 '''are''' distfiles. From §5.3.4 ''Checksum Phase Keywords''

     At least two checksum types (e.g., rmd160 and sha256) should be used
 to ensure the integrity of '''the distfiles'''.

 >  * The checksums line doesn't need to mention the distfile name, because
 there is only one file being checksummed.

 Ok.

 >  * Patchfile names should be of the form "patch-*.diff", as "`port lint
 --nitpick`" should mention.

 Fixed.

 >  * Because use you "`use_configure no`", you must add code to ensure
 you're UsingTheRightCompiler, build with `-arch` flags, and if possible
 add a universal variant.

 -- ''UsingTheRightCompiler''

 I have checked the Makefile and it uses ${CC}, I also added the
 recommended lines for ''Ports with nonstandard or non-existent configure
 scripts'' https://trac.macports.org/wiki/UsingTheRightCompiler
 #nonstandard-ports

 -- ''build with `-arch- flags`''

 Is this not handled by default?

 -- ''add a universal variant''

 Why? §5.3.7.1 ''Configure Universal''
     ''There is a default universal variant made available to all ports by
 MacPorts base, so redefining universal keywords '''should only be done'''
 to make a given port compile '''if''' the default options fail to do so.''

 >  * lib- and bin-style dependencies allow non-MacPorts software to
 satisfy them. We don't normally want that; see wiki:FAQ. So usually you
 should use port-style dependencies (if only a single port exists that can
 satisfy it), or path-style dependencies (if there are multiple ports that
 can satisfy it). Or if there is another reason why you used lib- and bin-
 style dependencies in this port, let me know.
 Switched both ''lib:'' and ''bin:'' dependencies to ''path:'' style. OS X
 is not supported by pwnt.org nor alternative package managers, so only
 macports alternative ''-devel'' ports will satisfy the dependencies in the
 foreseeable future.
 >
 > Regarding patch-Makefile:
 >
 >  * The install_name of a library should be the final location of the
 library when it is installed; it should not begin with ${DESTDIR}
 It has nothing to do with §7.2.4 ''DESTDIR: Support for Staged Installs''
 http://www.gnu.org/prep/standards/html_node/DESTDIR.html It's the smallest
 change I could think of to fix the dynamic library lookup path from
 relative to absolute. It's a nitpick, it works correctly either way, but
 since I'm fixing `port lint` trailing whitespaces...
 >  * The install_name should usually be set using the "`-install_name`"
 flag; is there a particular reason you used "`-Wl,-dylib_install_name,`"
 instead?
 `man ld` line 323-324.
  ''This option is also called -dylib_install_name for compatibility.''

-- 
Ticket URL: <https://trac.macports.org/ticket/43579#comment:5>
MacPorts <http://www.macports.org/>
Ports system for OS X