I've just committed a new version of the quartz device to R-devel. This seems to fix the problem with displaying math symbols in plots under Panther. I can't not more test R under Jaguar (at least not these days). In case you have the opportunity to do that, please send me a report if it brokes everything under Jaguar (or eventually say "it's ok") A couple of test as example(plotmath) and demo(plotmath) should suffice to see if things go smoothly. stefano
Math Symbols in plots under Darwin/Panther
4 messages · Simon Urbanek, stefano iacus
2 days later
On Feb 19, 2004, at 2:55 AM, stefano iacus wrote:
I've just committed a new version of the quartz device to R-devel. This seems to fix the problem with displaying math symbols in plots under Panther.
I gave it a shot on a Jag machine (darwin 6.8), here are my results.
I can't not more test R under Jaguar (at least not these days). In case you have the opportunity to do that, please send me a report if it brokes everything under Jaguar (or eventually say "it's ok")
Do we have the chicken-and-egg problem again? during the build, while building R_X11.so: ld: can't locate framework for: -framework R I thought this was fixed a while ago: you can't use -framework R when building R modules, because R is not installed as a framework yet. I had proposed the fake framework for building or different flags for initial build and packages build, but I don't remember which solution was used in the end - and it seems like none is used in the current R-devel :P The quick-and-dirty way to bypass the above is something like this: sudo bash # this will fail, but at least create the proper directories make install # we need to copy at least the library mkdir /Library/Frameworks/Versions/1.9.0/Resources/bin cp bin/libR.dylib /Library/Frameworks/Versions/1.9.0/Resources/bin cd /Library/Frameworks/R.framework/Versions ln -s 1.9.0 Current cd .. ln -s Versions/Current/Resources/bin/libR.dylib R now you should be able to re-run make this is just a hack for those who get stuck and want to test the R-devel anyway. Fixes for this problem are those discussed before. (I was away for a week so I didn't test R-devel in the meantime and I strongly suspect that this is not Jag-dependent)
A couple of test as example(plotmath) and demo(plotmath) should suffice to see if things go smoothly.
Well, define "smoothly" - on the Jag machine I tested the font face is constant and all math symbols and operators appear as boxes - i.e. the font is not changed, but no errors or crashes. Simon
On Feb 21, 2004, at 2:16 PM, Simon Urbanek wrote:
On Feb 19, 2004, at 2:55 AM, stefano iacus wrote:
I've just committed a new version of the quartz device to R-devel. This seems to fix the problem with displaying math symbols in plots under Panther.
I gave it a shot on a Jag machine (darwin 6.8), here are my results.
I can't not more test R under Jaguar (at least not these days). In case you have the opportunity to do that, please send me a report if it brokes everything under Jaguar (or eventually say "it's ok")
Do we have the chicken-and-egg problem again? during the build, while building R_X11.so: ld: can't locate framework for: -framework R
this should not happen. Now we only change this in the installed version of R not in build dir. Do you have another R in the path breaking up things?
I thought this was fixed a while ago: you can't use -framework R when building R modules, because R is not installed as a framework yet. I had proposed the fake framework for building or different flags for initial build and packages build, but I don't remember which solution was used in the end - and it seems like none is used in the current R-devel :P
what we do now is: * nothing in build dir of R, because of the problem above. * we changed etc/Makeconf in the R.framework only after R.framework installation could you please try to see what yo have in etc/Makeconf and bin/SHLIB inside your build dir? I mean, did you happen to have -framework R somewhere? ($LIBR) stefano
(I was away for a week so I didn't test R-devel in the meantime and I strongly suspect that this is not Jag-dependent)
ehm and surely you have done the build on a fresh (and cleaned) rysinced version of R-devel :)
A couple of test as example(plotmath) and demo(plotmath) should suffice to see if things go smoothly.
Well, define "smoothly" - on the Jag machine I tested the font face is constant and all math symbols and operators appear as boxes - i.e. the font is not changed, but no errors or crashes.
so, it doesn't work at all. This means that, that patch should not be applied on Jaguar and further means that I have to identify which version of OSX is running. thanks Simon, stefano
On Feb 21, 2004, at 2:16 PM, Simon Urbanek wrote:
On Feb 19, 2004, at 2:55 AM, stefano iacus wrote:
I've just committed a new version of the quartz device to R-devel. This seems to fix the problem with displaying math symbols in plots under Panther.
I gave it a shot on a Jag machine (darwin 6.8), here are my results.
A couple of test as example(plotmath) and demo(plotmath) should suffice to see if things go smoothly.
Well, define "smoothly" - on the Jag machine I tested the font face is constant and all math symbols and operators appear as boxes - i.e. the font is not changed, but no errors or crashes.
I've just committed a change. Under Jaguar the old behavior is maintained. Could you give another try? stefano