Re: What gets put in ~/.Mathematica and why?
- To: mathgroup at smc.vnet.net
- Subject: [mg122719] Re: What gets put in ~/.Mathematica and why?
- From: "Oleksandr Rasputinov" <oleksandr_rasputinov at hmamail.com>
- Date: Mon, 7 Nov 2011 05:54:00 -0500 (EST)
- Delivered-to: email@example.com
- References: <firstname.lastname@example.org>
On Sun, 06 Nov 2011 11:02:57 -0000, Mike Witt <msg2mw at gmail.com> wrote: > I just fooled around with CUDA on my system for the first time, and in > the process I got the message "Loading from Wolfram Research server" > and it appears that a *lot* of stuff was downloaded. As far as I can > tell it went into my *home* directory, into /home/mike/.Mathematica/ > Paclets. My .Mathematica directory is now, by far, the biggest thing > in my whole directory structure. > > I don't understand why this stuff didn't go into /usr/local/Wolfram. > Now this huge amount of stuff, is going to get backed up every time I > back up my personal directory. > > I'm curious if anyone can tell me what makes this (or any given > information) go into .Mathematica in my directory vs the "system" > place. > > If it makes a difference, I'm using Mathematica 220.127.116.11 Home Edition on > Fedora > 14 (32 bit) > > -Mike > The CUDA SDK is what is being downloaded. As you have seen, it is a sizeable package, but this is beyond WRI's control. As to why it wasn't stored in /usr/local/Wolfram: imagine a scenario where several users (logging into a cluster, say) are each doing CUDA development in the same environment. In this case, none of them may have write access to /usr/local. Also, each of them would surely prefer to update the SDK on their own terms, rather than risk having the behaviour of their codes change unexpectedly with unannounced updates to the CUDA headers and libraries. Of course, if you can think of a better location for the CUDA SDK on your own personal system, you're free to move it and create a symlink back to the expected location.