[R-sig-ME] lmer book draft

[Douglas Bates]

On Thu, Feb 18, 2010 at 3:30 PM, Rob Goedman <robjgoedman at me.com> wrote:
> Douglas,
>
> After my previous message I figured out how to work around the compile issue and
> have been able to get all 4 test scripts (for chapters 1, 2, 4 and 5) to work on MacOS
> SnowLeopard using 64 bit R.2.11 (but I think the same fix might also work on R.2.10,
> on Leopard and for 32 bit). It also required a later version of the Matrix package
> (available on R-forge).
>
> If someone needs help, they can just send me an email. It requires a temporary (just
> to install lme4a) change to a system file (alloca.h). I'm sure its not the final solution, but
> it seems to work for now.

Thanks Rob.  Indeed it is the problem with alloca.h that is hanging up
the compilation on Mac OSX.  The conflict is between a definition in
lme4utils.h, which does not have a prototype, and a definition in
header files downstream from the string header file, which does have a
prototype.

If you know an identifier for the compiler under Mac OSX that I can
use in an ifdef clause I will change that.  Not to worry if you don't,
I can check with Simon, or even grep through the R sources until I
find it.

An alternative, of course, it to remove that definition from a header
file that is included by C++ code.
> On Feb 18, 2010, at 12:47 PM, Douglas Bates wrote:
>
>> By the way, the Mac OSX version is failing to build for a different
>> reason.  It's another "twisty maze of include files" problem.
>>
>> On Thu, Feb 18, 2010 at 2:38 PM, Douglas Bates <bates at stat.wisc.edu> wrote:
>>> On Thu, Feb 18, 2010 at 2:19 PM, Andrew Dolman <andydolman at gmail.com> wrote:
>>>> Any chance of getting the compiled windows lme4a package? Or should I
>>>> expect to see it on r-forge very soon?
>>>
>>> I hope it will appear overnight on R-forge.  If not, I will follow up
>>> with the R-forge maintainers and, if necessary, make a compiled
>>> Windows binary from  win-builder available.
>>>
>>> I can see on the SVN archive that the lme4a sources are at revision
>>> 652.  (Windows compilations need revision 649 or later to succeed.) It
>>> looks like the last attempt to build a Windows version was around
>>> midnight CET yesterday, which is about 17:00 my time (CST).  I may
>>> have committed my fixes for Windows after that.  Right now the archive
>>> shows revision 649 as having been committed 23 hours ago.
>>>
>>>>>
>>>>> By the way, I have been able to build the lme4a package for Windows on
>>>>> Uwe's win-builder.r-project.org.  I checked the R-forge site for
>>>>> packages this morning but they haven't been re-built since last week.
>>>>> The latest SVN check-in that was built is revision 637 and the current
>>>>> revision is 650.
>>>>>
>>>>> _______________________________________________
>>>>> R-sig-mixed-models at r-project.org mailing list
>>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>>>>>
>>>>
>>>> _______________________________________________
>>>> R-sig-mixed-models at r-project.org mailing list
>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>>>>
>>>
>>
>> _______________________________________________
>> R-sig-mixed-models at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
>