TeXmacs has a feature called "sessions", where it can start up another program, like Octave, R or Maxima, and incorporate the input to and output from that program in the open document. Because of a bug in the way TeXmacs interfaces with R, this capability does not work for R versions 2.0.0 or later. I have notified the TeXmacs developers of this issue. However, I do not believe they have resolved it. My recommendation is to hold R at release 1.9.1 if TeXmacs is installed, until this bug is fixed.
Does the problem still persist in R-2.0.1?
Yes ... it is still there with R 2.0.1 and with the R 2.1 alpha test release. As far as I know it is not fixed in more recent TeXmacs releases either. It is a TeXmacs upstream problem. I'll go rattle the TeXmacs cages again and see if I can get them to fix it. The latest stable TeXmacs is 1.0.4-R3 and the latest development release is 1.0.4.7. Gentoo is still at 1.0.4.5. I'll test 1.0.4.7 from source and see if they've fixed it; my guess is they haven't. This bug is in the TeXmacs bug list already. See http://savannah.gnu.org/bugs/index.php?func=detailitem&item_id=10739
I just built TeXmacs 1.0.4.7 from source and it has the same problem. I'll go post an update on the TeXmacs bug list.
I haven't been able to get any attention on the TeXmacs list. Meanwhile, I took an R programming course in January and know how to work around this. I'll post the workaround to the TeXmacs lists, etc. and see what happens. Perhaps Gentoo can put in an "R" use flag in the TeXmacs ebuild, bring in R if the flag is set and activate the workaround at the end of the TeXmacs install.
Could you describe the workaround?
It's pretty hairy. I don't have the source directories handy at the moment, but what you have to do is go into the place where TeXmacs installs the R plugin. Then 1. Back up the original R plugin directory 2. Edit some files in the directory so they meet the R language standards. 3. Do "R CMD build" to make a source package from the edited directory. 4. Move the edited directory out of the way. 5. Do "R CMD INSTALL" to install the R package you just built. This installs a "working" R package where TeXmacs expects it to be. In short, its an ugly hack and a workaround for the real bug, which is the fact that TeXmacs distributes a binary package with its source and doesn't check for the presence or absence of R in the "configure" step. What needs to happen is 1. The developer of the R plugin for TeXmacs needs to read and understand the R standards for developing a source package. 2. Said developer needs to do the "R CMD build". This will create an R *source* package, which can be distributed with the TeXmacs source. 3. The TeXmacs install needs to check for the presence of R during the configure step and if it is present, do the "R CMD build" at the appropriate point in the installation. BTW, none of the above magic will work on Windows, although I'm told there are some hacks to the CygWin TeXmacs to make some pieces of it happen. R is not supported on CygWin, so it's not really worth the effort. If you guys want to attempt the hack in the TeXmacs ebuild, let me know, but I'd settle for "RESOLVED UPSTREAM".
This bug has been fixed in TeXmacs-1.0.6.6. The ebuild can be found at http://gentooscience.org
As I said, this bug has been fixed long ago. Anybody interested in running R, or recent versions of maxima, from TeXmacs, should install a recent version of TeXmacs from Gentoo science overlay.
This finally got fixed upstream, but I don't remember which release of TeXmacs it was in. Go ahead and close it.
Closing per reporter request.
