Github: alt.sources.* rides again, on steroids.

It’s back to he days of alt.sources.*.   With the advent of github, I seem to be building my favorite tools  from sources again most of the time.   They work better, faster with no waiting on packagers or vendors.

For something I was working on, I wound up having to clone part of a git tree into a new repository, which turned out to be surprisingly hard. This helped:

http://ariya.ofilabs.com/2014/07/extracting-parts-of-git-repository-and-keeping-the-history.html

But it lead to  pulling the source for for git itself to have git-subtree work

http://engineeredweb.com/blog/how-to-install-git-subtree/

and I’ve already decided that building  with the demise of Damien Cassou’s emacs snapshots https://launchpad.net/~cassou/+archive/ubuntu/emacs, the best route is to build emacs from the latest sources https://github.com/eludom/HOWTO/blob/master/emacsFromSrc.org, and for various reasons the same goes for org-mode https://github.com/eludom/HOWTO/blob/master/getLatestOrg.sh

Things seem to have come full circle.   Back before RPM and apt-get we used to pull the latest source out of the local alt.sources.* Usenet archives and build everything (to this day I have a ~/src/ or ~/build/ directory on many  systems, but it’s migrating to ~/git/${GITHUB_USERNAME}/package

When an emacs user calls something aRcane…

I’m working through an online course on the R language https://www.coursera.org/course/rprog

I’m finding it quite arcane, retro, baroque. Just a few examples:

  • length(x)=5, assigning to a function to to set the length of an object? really ?
  • “1” based array index (FORTRAN, anyone?)
  • complete textual orientation (OK, not a problem for an emacs user)
  • Having to think about the size of objects being read in (echos of the days before virtual memory)

On the other hand, I was able to generate a heatmap in 3 lines of code (thanks to Bronwyn Woods for the pointer). I spent hours messing around with Python and JavaScript versions, but the R version Just Worked. Three lines. Amazing.

data <- data.frame(x=c(1,2,3), y=c(1,2,3), count=c(10,10,5))
library(lattice)
levelplot(count ~ x*y, data=data)

Rplots.jpg

Figure 1: Simple heatmap generated by R