Problem with readline when building R from source.
On 24/05/2010, at 6:59 PM, Prof Brian Ripley wrote:
<SNIP>
....... I actually use tcsh, so I did the equivalent tcsh thing: setenv F77 "gfortran -arch x86_64" setenv FC "gfortran -arch x86_64"
But the documented way is to put these in a config.site file: and yes, it is in the R-admin manual. So all that was needed was to put in file config.site (in the top-level R build directory) CC='gcc -arch x86_64' CXX='g++ -arch x86_64' F77='gfortran -arch x86_64' FC='gfortran -arch x86_64' OBJC='gcc -arch x86_64'
(copied from the manual).
OK. Got that. Put it in the config.site file. It seemed to fly all right.
That solved the fortran problem, and the configure goes on well past the point where it hung up last time. But it eventually comes to a shuddering halt with an error about X11 --- which is where I came in, really. I'm toadally mystified. The error I'm getting is: configure: error: --with-x=yes (default) and X11 headers/libs are not available That is just like the error I was getting in respect of readline; I was told to update my readline (via a fancy ``curl'' command). I looked in //r.research.att.com/libs and could not see any X11 package to which to apply a curly.
This one is probably that mentioned in the the R-admin manual for Snow
Leopard, to wit
'Another quirk is that people have found that the X11 libraries are
not in the default linking path, so something like
@samp{LIBS=-L/usr/X11/lib} may be required.'
So add that line to the config.site file.
Did so, and *that* unfortunately did not seem to work. I'm still getting the
configure: error: --with-x=yes (default) and X11 headers/libs are not available
error, and configure bails out at that stage. Can you suggest something else that I might try, or suggest ways in which I could track down where the appropriate X11 material lives and how I can pass this information on to ``configure''? I don't know what the X11 ``headers/libs'' files are called, or what they look like. Up to now I've just treated it all as a black box, and up to now that's worked. I did a ``locate X11'' just now and got over 14 ***thousand*** lines of output. Bozhemoi!
What the <expletive deleted> do I do now? (And why do these things always happen to ***me***???)
I think the foregoing bears repeating.
Perhaps because you don't study the manuals? I've yet to see anything in this thread that is not covered in the general advice and the Mac OS X section of the R-admin manual.
Seriously Brian, it's all very well to say ``read the manual'', and that works when one has a fairly solid understanding of what's ***in*** the manual and one just needs to check on a detail of syntax, or like that. But when one hasn't a clue as to what the underlying mechanisms are, that are triggering the problem, then reading the manual could entail a week's hard work with no guarantee at all of any positive results. It might be the case that the error is being triggered by some other quirk of the OS that actually has nothing to do with R. If one is sufficiently knowledgeable, one can discern this. If one is not, it's all a big mystery. Actually I wouldn't even know *which* manual to start looking in. Now that I know it's the R-admin manual, it's obvious. But things are [almost] always obvious once you know them.
We can't spell out the exact details for every possible system -- the possibilities are too many and the info would go out of date too rapidly. But for a '0 new features' release, Snow Leopard sure did introduce a whole raft of barely documented changes and we've not yet caught up with some in configure (ideally X11 would be looked for in /usr/X11/lib, but what was Apple thinking of in dropping it?).
What indeed? cheers, Rolf ###################################################################### Attention: This e-mail message is privileged and confidential. If you are not the intended recipient please delete the message and notify the sender. Any views or opinions presented are solely those of the author. This e-mail has been scanned and cleared by MailMarshal www.marshalsoftware.com ######################################################################