Dear Pierre, Jean-François and others,
I read with interest your discussion on GCDkit. Concerrning the
litany on its installation, I stress that we have tried our best to make
it simple. However, as one of the Murphy laws reads, you can make it
foolproof but you cannot make it damnfoolproof. ;-)
Doesn't the profiler allows to select several columns and plot them together ?Yes, it does, and it does allow formulae at the moment when the variables are being specified. So, for instance,you can write SiO2/10, MgO*2 and TiO2*10 if you please.
And also, I did not find the way to adjust scales between various elements.
Yes: I was just wandering whether there was not something already prepared for, say, hydrothermal alterations, metasomatisms, stuff like that.I do not work very much with such problems, so the system does not have too many tools. It does contain a plugin for isocon plots (after Gresens) plus one can also use the spiderplots, as it is possible to normalize by selected sample or average of some of them.
Yes, I guess so. But I'll rather do it from my R on my linux box (R version 2.11.0 (2010-04-22)), to get all the comfort of a true command-line environment.I think the real power, especially for novice users, is to actually use the combination of command line and GUI.
Actually, I'm more used to do my plots with gnuplot, so learning R to do such simple things as plots is a bit chiasseur...I would say, that some plots can become rather complex. And that GCDkit allows quite flexible functions for postprocessing the plots and using them e.g. for classification.
But, to go back to GCDkit: are there any plans to make it multiplatform? It seems that most is in Tk, so it should not be too difficult to port, at first glance.Yes, for several years already, but I am very slow. The trouble is that the menus and dialogue boxes are not really Tk but are all windows specific. They are the only reason why we stick to Windows for the moment. First of all, we have to remove all the hardwired dialogues and redesign the functions so they could be called from the command line (this is at rather advanced stage now). The next step will be to develop a set of interfaces, one of them most likely in Tcl/Tk. The other possibilities such a design will open is to run GCDkit in batch mode, from command line, or even returning the results to Latex or Open Office documents/reports. I have experimented with this, too, with some interesting results.