[Rd] R-devel on FreeBSD: Support for C99 complex type is required

[Rainer Hurling]

On 08.02.2011 22:10 (UTC+1), Prof Brian Ripley wrote:
> On Tue, 8 Feb 2011, Rainer Hurling wrote:
>> On 07.02.2011 23:45 (UTC+1), Prof Brian Ripley wrote:
>>> A follow-up on this.
>>>
>>> Cygwin has recently[*] added support for C99 complex math, taken from
>>> NetBSD with code that is very similar to that from Steven Moshier
>>> available via http://www.moshier.net/c9x_readme.html.
>>>
>>> That code isn't entirely right, especially not at the cuts on the
>>> inverse functions where C99 mandates what cut is used (and neither glibc
>>> nor Mac OS X have it correct).
>>>
>>> I would expect that the current R-devel (I mean an SVN checkout now)
>>> should build on your platform, substituting the missing functions by
>>> ones based on earlier code. There may be a few more tweaks required, but
>>> I have corrected several errors in the versions FreeBSD would have used
>>> in R 2.12.1.
>>
>> On FreeBSD 9.0-CURRENT (amd64) I have done a
>>
>> svn co https://svn.r-project.org/R/trunk r-devel/R
>>
>> and after that
>>
>> configure --without-recommended-packages
>>
>> The configure script does not complain any more about missing C99
>> complex support. I was able to build and install R (without
>> recommended packages).
>>
>> Starting R and doing examples on ?complex seems to work correct.
>>
>> sessionInfo()
>> R version 2.13.0 Under development (unstable) (2011-02-08 r54279)
>> Platform: x86_64-unknown-freebsd9.0 (64-bit)
>> locale:
>> [1]
>> de_DE.ISO8859-15/de_DE.ISO8859-15/C/C/de_DE.ISO8859-15/de_DE.ISO8859-15
>> attached base packages:
>> [1] stats graphics grDevices utils datasets methods base
>> loaded via a namespace (and not attached):
>> [1] tools_2.13.0
>>
>> Thank you very much for this work. Will it exist in some way in the
>> next release?
>
> Just to be clear, the next release will almost certainly be 2.12.2
> (currently R-patched). That will use pre-C99 complex as 2.12.1 did, but
> with several bugs fixed. Then we would expect a 2.13.0 in April, and
> that will require C99 complex in the compiler plus some version of the
> current substitutes for csin etc.

Sorry, my last question was imprecise. Of course I meant 2.13.x.

> My sysadmins have updated a FreeBSD virtual machine for me (to 8.2-rc3):
> hopefully I will be able to test R-devel on it in future.

Wow, that's interesting. Please let me know if I could test something on 
FreeBSD 9.0-CURRENT (the amd64 development version).