[Phylobase-devl] various build questions
Mark Holder
mtholder at ku.edu
Fri Jul 23 00:59:52 CEST 2010
Hi,
I got it to build on my windows machine. I posted the files that I had to change at:
http://phylo.bio.ku.edu/content/ncl-fixes-were-breaking-windows-64-bit-build-through-mingw-g-42
Alternatively the project admins could give me svn write access if you want me to commit the changes.
all the best,
Mark
On Jul 22, 2010, at 12:10 PM, Brian O'Meara wrote:
> Thanks, Mark. One way to look at CRAN's checks for its release of phylobase is http://cran.r-project.org/web/checks/check_results_phylobase.html
>
> And R-forge's info on builds for the latest version is at https://r-forge.r-project.org/R/?group_id=111 (but you have to look into each daily build/daily check log).
>
> CRAN shows v 0.5 builds on everything but Solaris, but R-forge shows v 0.6 has errors on Windows 32 bit and Mac 32 bit.
>
> There may be better ways of checking all this.
>
> Brian
>
> On Jul 22, 2010, at 12:29 PM, Mark Holder wrote:
>
>> Hi again,
>> http://win-builder.r-project.org/aJTkLv5eh3pA/00install.out
>>
>> is now returning a 404.
>>
>>
>> Do we know that the phylobase windows build is still broken?
>>
>>
>> (sorry for the newbie question, I don't normally use R so I'm pretty unfamiliar with the R community and the basics for where I should look for such info).
>>
>>
>> Mark
>>
>>
>>
>> On Jul 22, 2010, at 11:24 AM, Mark Holder wrote:
>>
>>> Hi Ben,
>>> Just a status report.
>>>
>>>
>>> As it turns out, it is not so easy to build R on windows (particularly for those of us who had previous CygWin installations).
>>>
>>> The link labelled
>>> mingw-w64-1.0-bin_i686-mingw_20100322.zip
>>> was in fact just a link to the MinGW-64 sourceforge site. That particularly version of the mingw-w64-1.0-bin_i686-mingw tools does not seem to be available anymore.
>>>
>>> I notice that today the link has been changed to say "R 2.11.0 used http://sourceforge.net/projects/mingw-w64/files/Toolchains%20targetting%20Win64/Automated%20Builds/mingw-w64-1.0-bin_i686-mingw_20100322.zip, but this is apparently no longer available for download." I guess that it was just bad luck that I tried to do this the for the first time when they are just discovering that the toolkit they require is no longer available for download.
>>>
>>> The more recent version of mingw-w64-1.0-bin_i686-mingw_.... from http://sourceforge.net/projects/mingw-w64/files/Toolchains%20targetting%20Win64/Automated%20Builds/
>>> include gcc 4.5 This does not appear to be compatible with R-2.11.x
>>>
>>> I'm currently investigating use the snapshots found at http://www.stats.ox.ac.uk/pub/Rtools/ to build.
>>>
>>> But it is all a bit complicated because that page mixes and matches build tools for the eight windows build variant. 8 comes from 2^3 where the 3 are the three binary choices for:
>>> - gcc 4.5 vs earlier
>>> - R2.11. vs R2.12
>>> - 32 bit vs 64 bit
>>>
>>> Based on http://win-builder.r-project.org I think I just need R-2.11.1 on 64bit. I think that means that I have to use gcc 4.2.
>>>
>>>
>>>
>>>
>>> For what it is worth:
>>>
>>> The reason that the newest gcc breaks R build (for me) is similar to what is described at http://tolstoy.newcastle.edu.au/R/e11/devel/10/07/0032.html except that I see the problems during linking not at the make check stage. When deciding what symbols to include in a DLL, R's builds have rules that used nm to list all symbols, then they use sed to remove underscores from the names (assuming that g++ will mangle names in that way exactly). I guess that g++ changed their name mangling.
>>>
>>> I confess that I do not understand R's make configuration system. I've never seen one like it. I don't want to sound hypercritical, but doing things like piping nm to sed to figure out what symbols you want to include is not a path to long term happiness in life ;-)
>>>
>>>
>>>
>>> I'll let you know if I make any progress.
>>>
>>>
>>> Mark
>>>
>>>
>>>
>>> On Jul 18, 2010, at 10:25 AM, Ben Bolker wrote:
>>>
>>>> thanks Mark.
>>>>
>>>> Mark Holder wrote:
>>>>> Hi,
>>>>> I'll give it another shot. It'll probably be Wednesday before I even have a chance to try.
>>>>>
>>>>> all the best,
>>>>> mark
>>>>>
>>>>>
>>>>> On Jul 15, 2010, at 9:19 AM, Ben Bolker wrote:
>>>>>
>>>>>> About 64-bit toolchain:
>>>>>> http://www.murdoch-sutherland.com/Rtools/ says:
>>>>>>
>>>>>>
>>>>>> Tools for 64 bit Windows builds
>>>>>>
>>>>>> Most of the tools used for 32 bit builds work fine as well for 64 bit
>>>>>> builds, but the gcc version is necessarily different. We use the
>>>>>> MinGW-w64 version, available from SourceForge.net.
>>>>>> R 2.11.0 used mingw-w64-1.0-bin_i686-mingw_20100322.zip.
>>>>>> Current snapshot builds are currently using the same version, but we may
>>>>>> update before the next release of R.
>>>>>>
>>>>>> To install, simply unzip the whole file into a directory such as
>>>>>> Rtools/MinGW-w64, and add Rools/MinGW-w64/bin to your PATH ahead of the
>>>>>> Rtools/MinGW/bin directory. Then follow the instructions in the "R
>>>>>> Administration and Installation" manual, section "Building R for 64-bit
>>>>>> Windows" to build R. Building packages proceeds as on other platforms
>>>>>> provided 64 bit R and the 64 bit gcc are found on the PATH before 32 bit
>>>>>> versions. Note that these instructions are currently evolving rapidly;
>>>>>> plans are for a single combined build by the time 2.12.0 is released.
>>>>>>
>>>>>> That seems fairly unambiguous?
>>>>>> Are you willing to take a stab at installing those tools and see if
>>>>>> it helps replicate the problem?
>>>>>>
>>>>>> cheers
>>>>>> Ben
>>>>>>
>>>>>>
>>>>>>
>>>>>> Mark Holder wrote:
>>>>>>> Hi,
>>>>>>> I've just been playing around with the Windows builds. I'm afraid that I have very little to report.
>>>>>>>
>>>>>>> It is hard for me to figure out exactly what version MinGW I should be using to replicate this (and there are issues with the MinGW version that I selected, making it pretty useless right now). Does anyone know the procedure for getting a 64bit machine up and building C-extensions to R?
>>>>>>>
>>>>>>> I just committed a change to:
>>>>>>> https://ncl.svn.sourceforge.net/svnroot/ncl/branches/v2.1
>>>>>>> that allows it to build on a 64 bit machine using visual studio 2010. But that change seems to be totally unrelated to the g++ issues that show up on http://win-builder.r-project.org/
>>>>>>>
>>>>>>>
>>>>>>> I'll see if I can install a bleeding edge g++ on one of my linux boxes. Perhaps that will reveal the same complaints and I can diagnose.
>>>>>>>
>>>>>>> NCL's "nxsstring" class uses some hacky operator overloading which seems to be the source of the problem, but I'm not sure of the exact cause. That code has been compiling for years... the joys of working with C++.
>>>>>>>
>>>>>>> I'll let you know if I can figure out anything. If anyone wants an account on my windows machine to work on this remotely, I can set up ssh on it and get you an account set up.
>>>>>>>
>>>>>>>
>>>>>>> Mark
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Jul 8, 2010, at 4:58 PM, Ben Bolker wrote:
>>>>>>>
>>>>>>>> hi folks ...
>>>>>>>>
>>>>>>>> current version of phylobase is building/checking on my machine
>>>>>>>> (32-bit Linux: it was choking on R CMD check, but re-installing seems to
>>>>>>>> have fixed that).
>>>>>>>>
>>>>>>>> * R-forge shows that builds on R-devel (2.12.0) are not working because
>>>>>>>> "Rscript" can't be found (this applies both to linux and windows 32 bit).
>>>>>>>>
>>>>>>>> https://r-forge.r-project.org/R/?group_id=111&log=build_win32&pkg=phylobase&flavor=devel
>>>>>>>>
>>>>>>>> I'm having some trouble building R-devel on my machine but will
>>>>>>>> hopefully succeed soon.
>>>>>>>>
>>>>>>>> * Furthermore, 64-bit windows builds on win-builder.r-project.org are
>>>>>>>> failing, with a variety of errors, listed in:
>>>>>>>>
>>>>>>>> http://win-builder.r-project.org/aJTkLv5eh3pA/00install.out
>>>>>>>>
>>>>>>>> Any thoughts? Anyone have a 64-bit windows machine for testing
>>>>>>>> purposes ... ? (I will see if I can get access to one here, but it
>>>>>>>> could take a little while ...)
>>>>>>>>
>>>>>>>> cheers
>>>>>>>> Ben
>>>>>>>>
>>>>>>>> the first few errors in the 64-bit windows build are:
>>>>>>>>
>>>>>>>> nxsmultiformat.cpp: In member function 'bool
>>>>>>>> MultiFormatReader::readFastaSequences(FileToCharBuffer&, const
>>>>>>>> NxsDiscreteDatatypeMapper&, std::list<std::basic_string<char,
>>>>>>>> std::char_traits<char>, std::allocator<char> >,
>>>>>>>> std::allocator<std::basic_string<char, std::char_traits<char>,
>>>>>>>> std::allocator<char> > > >&, std::list<std::vector<int,
>>>>>>>> std::allocator<int> >, std::allocator<std::vector<int,
>>>>>>>> std::allocator<int> > > >&, size_t&)':
>>>>>>>> nxsmultiformat.cpp:334: error: ambiguous overload for 'operator<<' in
>>>>>>>> '((NxsString*)((NxsString*)err.NxsString::operator<<(((const
>>>>>>>> char*)"Illegal state code
>>>>>>>> \"")))->NxsString::operator<<(((int)c)))->NxsString::operator<<(((const
>>>>>>>> char*)"\" found when reading character ")) <<
>>>>>>>> ((NxsDiscreteStateRow*)row)->std::vector<_Tp, _Alloc>::size [with _Tp =
>>>>>>>> int, _Alloc = std::allocator<int>]()'
>>>>>>>> ./ncl/nxsstring.h:582: note: candidates are: NxsString&
>>>>>>>> NxsString::operator<<(int)
>>>>>>>> ./ncl/nxsstring.h:591: note: NxsString&
>>>>>>>> NxsString::operator<<(unsigned int)
>>>>>>>> ./ncl/nxsstring.h:600: note: NxsString&
>>>>>>>> NxsString::operator<<(long int)
>>>>>>>> ./ncl/nxsstring.h:609: note: NxsString&
>>>>>>>> NxsString::operator<<(long unsigned int)
>>>>>>>> ./ncl/nxsstring.h:618: note: NxsString&
>>>>>>>> NxsString::operator<<(double)
>>>>>>>> ./ncl/nxsstring.h:627: note: NxsString&
>>>>>>>> NxsString::operator<<(const char*) <near match>
>>>>>>>> ./ncl/nxsstring.h:636: note: NxsString&
>>>>>>>> NxsString::operator<<(char)
>>>>>>>> ./ncl/nxsstring.h:645: note: NxsString&
>>>>>>>> NxsString::operator<<(const std::string&) <near match>
>>>>>>>> ./ncl/nxsstring.h:544: note: NxsString&
>>>>>>>> NxsString::operator<<(NxsString& (*)(NxsString&)) <near match>
>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>>> Ben Bolker
>>>>>>>> *** NEW E-MAIL ADDRESSES:
>>>>>>>> *** bbolker at gmail.com , bolker at math.mcmaster.ca
>>>>>>>> bolker at ufl.edu / people.biology.ufl.edu/bolker
>>>>>>>> GPG key: people.biology.ufl.edu/bolker/benbolker-publickey.asc
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> Phylobase-devl mailing list
>>>>>>>> Phylobase-devl at lists.r-forge.r-project.org
>>>>>>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/phylobase-devl
>>>>>>
>>>>>> --
>>>>>> Ben Bolker
>>>>>> *** NEW E-MAIL ADDRESSES:
>>>>>> *** bbolker at gmail.com , bolker at math.mcmaster.ca
>>>>>> bolker at ufl.edu / people.biology.ufl.edu/bolker
>>>>>> GPG key: people.biology.ufl.edu/bolker/benbolker-publickey.asc
>>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Ben Bolker
>>>> *** NEW E-MAIL ADDRESSES:
>>>> *** bbolker at gmail.com , bolker at math.mcmaster.ca
>>>> bolker at ufl.edu / people.biology.ufl.edu/bolker
>>>> GPG key: people.biology.ufl.edu/bolker/benbolker-publickey.asc
>>>>
>>>
>>
>> Mark Holder
>>
>> mtholder at ku.edu
>> http://phylo.bio.ku.edu/mark-holder
>>
>> ==============================================
>> Department of Ecology and Evolutionary Biology
>> University of Kansas
>> 6031 Haworth Hall
>> 1200 Sunnyside Avenue
>> Lawrence, Kansas 66045
>>
>> lab phone: 785.864.5789
>>
>> fax (shared): 785.864.5860
>> ==============================================
>>
>>
>>
>>
>>
>>
>>
>>
>> _______________________________________________
>> Phylobase-devl mailing list
>> Phylobase-devl at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/phylobase-devl
>
> ------------------------------------------------------
> Brian O'Meara
> http://www.brianomeara.info
> Assistant Prof.
> Dept. Ecology & Evolutionary Biology
> U. of Tennessee, Knoxville
>
Mark Holder
mtholder at ku.edu
http://phylo.bio.ku.edu/mark-holder
==============================================
Department of Ecology and Evolutionary Biology
University of Kansas
6031 Haworth Hall
1200 Sunnyside Avenue
Lawrence, Kansas 66045
lab phone: 785.864.5789
fax (shared): 785.864.5860
==============================================
More information about the Phylobase-devl
mailing list