[46233] trunk/dports/math

[Ryan Schmidt]

On Feb 1, 2009, at 02:36, Juan Manuel Palacios wrote:

> 	Hey Ryan! A quick favor I wanted to ask you, if I may.
>
> 	I just added the Portfile below to the repo, but my Portfile  
> writing skills are a bit rusty so I wanted to ask you, if I may, to  
> please give it a look and let me know if I made any obvious boo- 
> boos or if there's anything in it that could be achieved in a  
> simpler way.

Of course! I was just about to anyway. :)

> Begin forwarded message:
>
>> From: jmpp at macports.org
>> Date: February 1, 2009 3:59:16 AM GMT-04:30
>> To: macports-changes at lists.macosforge.org
>> Subject: [46233] trunk/dports/math
>> Reply-To: macports-dev at lists.macosforge.org, jmpp at macports.org
>>
>> Revision
>> 46233
>> Author
>> jmpp at macports.org
>> Date
>> 2009-02-01 00:29:04 -0800 (Sun, 01 Feb 2009)
>> Log Message
>>
>> New port: mapm 4.9.5, (yet another) arbitrary precision math  
>> library in C.
>> Added Paths
>>
>> trunk/dports/math/mapm/
>> trunk/dports/math/mapm/Portfile
>> Diff
>>
>> Added: trunk/dports/math/mapm/Portfile (0 => 46233)
>> --- trunk/dports/math/mapm/Portfile	                        (rev 0)
>> +++ trunk/dports/math/mapm/Portfile	2009-02-01 08:29:04 UTC (rev  
>> 46233)
>> @@ -0,0 +1,39 @@
>> +# $Id$
>> +
>> +PortSystem 1.0
>> +
>> +name    mapm
>> +version    4.9.5
>> +categories    math science
>> +platforms	darwin
>> +maintainers	jmpp at opendarwin.org
>> +
>> +description    A Portable Arbitrary Precision Math Library in C.
>> +long_description    MAMP is a set of functions that allow the user \
>> +                    to perform math to any level of accuracy that  
>> is desired. \
>> +                    The precision of a number is only limited by  
>> 'INT_MAX' \
>> +                    and available memory.
>> +
>> +homepage    http://www.tc.umn.edu/~ringx004/mapm-main.html
>> +master_sites    http://www.tc.umn.edu/~ringx004
>> +checksums    md5 a92ff86b6240f78f882661e0b1c11920 \
>> +             sha1 99dab13404ee30e7a830204da0a66cbae265a8b5 \
>> +             rmd160 f4e39962410ea4f9a0d85a3e07b7421e9795d6cb
>> +
>> +use_configure    no
>> +
>> +worksrcdir   ${name}_${version}
>> +build.cmd    make -f makefile.osx

I think this would work better if you replace the build.cmd line with:

build.args -f makefile.osx

because we currently have the bug/feature that if build.cmd has a  
space in it, then we never attempt to run the command through "nice";  
that was the "fix" for #16091:

http://trac.macports.org/ticket/16091

>> +destroot {
>> +    copy ${worksrcpath}/m_apm.h ${destroot}${prefix}/include
>> +    copy ${worksrcpath}/libmapm.a ${destroot}${prefix}/lib
>> +    xinstall -d -m 755 ${destroot}${prefix}/share/${name}
>> +    copy ${worksrcpath}/DOCS ${destroot}${prefix}/share/${name}/
>> +    copy ${worksrcpath}/MULTI_THREAD ${destroot}${prefix}/share/$ 
>> {name}/
>> +    copy ${worksrcpath}/PI_DEMO ${destroot}${prefix}/share/${name}/
>> +    copy ${worksrcpath}/README ${destroot}${prefix}/share/${name}
>> +    foreach prog  {calc primenum validate} {
>> +        copy ${worksrcpath}/$prog ${destroot}${prefix}/bin
>> +    }
>> +}


Docs should go in ${prefix}/share/doc/${name} (or perhaps ${prefix}/ 
share/doc/${name}-${version} -- there was a push for that last year  
(or before?) but I'm not sure the justification was totally sound).

I tend to use "xinstall" in the destroot phase and "copy" in other  
phases. There's not much difference, I think, in the outcome. But  
using xinstall you can simplify the above to:

destroot {
	xinstall -m 755 -W ${worksrcpath} \
		calc primenum validate \
		${destroot}${prefix}/bin
	xinstall -m 644 -W ${worksrcpath} m_apm.h ${destroot}${prefix}/include
	xinstall -m 644 -W ${worksrcpath} libmapm.a ${destroot}${prefix}/lib
	xinstall -d -m 755 ${destroot}${prefix}/share/doc/${name}-${version}
	xinstall -m 644 -W ${worksrcpath} \
		DOCS MULTI_THREAD PI_DEMO README \
		${destroot}${prefix}/share/doc/${name}-${version}
}

(Not tested; typos possible...)

If you do this change, since it causes the docs to go to a different  
place, the revision should be incremented.