BTW - I should add that my memory is that there is at least one package for
having R interact with Python that does an Anaconda install without you really
knowing it. Personally I consider that bad practice. It may be more
convenient but can lead to unexpected trouble down the line just becau
You are welcome (my one good deed for the day - I guess I can go to bed
now!!!) Anaconda is a great product, I use it for my Python work, they
actually ask you if you want them to modify the .profile, but people say yes
without realizing the implications. First time I installed it, I said y
Roy, thanks for this! Your insight explains a great deal of my
problems. And indeed I see a whole bunch of PATH additions which were
created by anaconda (these are in .bash_profile, certainly not put
there by me!).
Just to clarify, I'm not using anaconda by choice, it seems to have
been install
Anaconda adds code to the end of your ,.profile that can really play havoc with
paths in other programs/settings - and off I remember correctly it puts the
changes at the front of your $PATH.. I moved that code in .profile to a
separate file. When I want anaconda, I just source the file. Mak
That's anaconda - they don't include paths which is why it breaks everything
until you insert a manual override via DYLD_LIBRARY_PATH. See Robin's e-mail he
is NOT using the system curl, but rather a custom version from anaconda
("/Users/rhankin/anaconda3/lib").
Cheers,
Simon
> On Feb 25, 20
If your goal is to interoperate with Python (esp. when using numpy and friends)
then Anaconda is actually a good option... just learn how activate a conda
environment before doing anything within that ecosystem and remember that each
conda environment you create will be independent of all others
Simon, thanks for this. Can we add your comment to the mac os x FAQ
please?What is the recommended way to install libcurl?
My feeling is that I have somehow messed up my system so badly that
reimaging is the only way forward.
hankin.ro...@gmail.com
hankin.ro...@gmail.com
On Tue, Feb 2
You're apparently using anaconda - it means you have to activate it first
because it doesn't include paths in its binaries so ti won't work without
explicit path override. Also note that anaconda is by design incompatible with
the system libraries, so you'll have to make sure everything you use
thanks Peter, my results are:
rstudio % ls -l /usr/lib/*curl*
lrwxr-xr-x 1 root wheel 15 4 Dec 2017
/usr/lib/libcurl.3.dylib -> libcurl.4.dylib
-rwxr-xr-x 1 root wheel 854416 4 Jul 2018 /usr/lib/libcurl.4.dylib
lrwxr-xr-x 1 root wheel 15 4 Dec 2017 /usr/lib/libcurl.dylib
->
Hmm, how did "@rpath" get in there? I don't see it anywhere in my build and
source dirs. Looks like something that should have been substituted (by perl?)
with the appropriate path.
Is your library (...anaconda3/lib) correctly installed? I have, for the system
libcurl:
pd$ ls -l /usr/lib/*curl
thanks for this Peter, my config.log file looks like this:
configure:42615: result: yes
configure:42629: checking if libcurl is version 7 and >= 7.22.0
configure:42658: gcc -o conftest -g -O2
-I/Users/rhankin/anaconda3/include -I/usr/local/include
-I/usr/local/include -L/usr/local/lib conftest.c
It doesn't usually happen on 10.13.6... Best guess is that somehow you are
picking up an older library (e.g. in /usr/local/lib) even if you are using
curl-config from a newer version.
The detective work needed could take off from config.log, region around this:
configure:39937: checking if libc
12 matches
Mail list logo
