7:30 – 5:00 VISIBILITY

• trying to get the fgmdev version of R matching the dev version
• everything but vcd and HH are in the debian distribution
  • installed and loaded vcd from within R
  • HH failed. It needs a newer version of multcomp. Grr. Re-rigged the code so that it’s not needed.
• Adding running totals of coefficients to ANCOVA – done
• Checked up on PPM, which is running fine, but not being used. Helped write up more justification with Bill. Noticed that Matt H. was poking around in some of the + data.
• Reworked the creation of tables so that they can have columns with different numbers of items in them. Did this by inserting the special ‘NA’ symbol into the list after the last element up to the length of the longest element.

7:30 – 5:30

• Hey! It’s Spring! Boy, is it wet.
• Tried to connect Outlook to the FGM webmail, just to see if that could work. So far, it just seems to kill Outlook.
• Adding image size arguments to evaltoJpeg() – done
• Adding a “Save Chart” option – done
• Deleted and reloaded R on fgmdev. I have a script for removing and installing components, since install.packages() seems to break the installation. Did a quick test by running library(multcomp), which did not complain. Will try Rserve later.
• Adding running totals of coefficients to ANCOVA
• Making sure that the coefficient order matches the group order
• Adding handling of NA in tables and/or dataframes
• Need to figure out how to get multcomp working on debian – done. Next is HH

7:30 – 7:00 VISIBILITY

• Testing code today
• Need to add image size arguments to evaltoJpeg()
• Need to add running totals of coefficients to ANCOVA
• Need to make sure that the coeeficient order matches the group order
• Need to add handling of NA in tables and/or dataframes

8:00 – 6:30 VISIBILITY

• CRAN is down, which is messing up my testing – it’s back up again. Test tomorrow.
• Need to write up a progress report for Larry
• Spent the rest of the day putting together complex contrasts, which required reworking the RTestBase class, which then requred fixes of the Student T and ANOVA classes. Hopefully, I can get all of this working on fgmdev, I was having some trouble with library(HH) before CRAN went down.

7:30 – 6:00 VISIBILITY

• Caught up with Dong a bit on the Financial Status part that we were working on before the email fire drill
• We also discussed generating VISIBILITY reports based on financial data. I’ll be showing him how the Scripting system works so he can start mining the PPM data.
• Scatterplots and coefficients and CONOVA, oh my!
• And I need to tie into the VISIBILITY remote object so I can start working with those tables…
• CMM meeting. They like what they see, but don’t want to pay for it.
• More privac crap

7:30 – 4:30 VISIBILITY

• Trying to get R installed on fgmdev.
  • Followed the directions here (http://www.mayin.org/ajayshah/KB/R/documents/install.html). The packages appear to have downloaded and installed, but I am unable to get the console running
  • Ah. I needed to run ‘apt-get install r-base’
  • Installed the following packages: Rserve, car, vcd, effects, gplots, multcomp.
  • set R_HOME to /usr/lib/R
  • set LD_LIBRARY_PATH to $R_HOME/lib
  • exported LD_LIBRARY_PATH
  • ran Rserve, which started Rserve as a daemon.
• Meeting with Bill to discuss server things. I told him that I had been in contact with Denise WRT getting privac. He said that he would follow up, also got a note from Kris U. about tying into a mail server. Looks very doable.

7:30 – 3:30 visibility

• VISIBILITY is back up on site
• Denise responded to my privac request and passed it off to someone else in her office
• Kris asked about if we required certs in our app. I told him no, that we could do without them.
• Worked on adding Kruskal-Wallis and Welche’s W test to the one way ANOVA widget. I’m going to make it so that you check the tests and plots you want
• Need to get R up and running on FGMdev so I can run this facing out.

7:30 – 6:30 VISIBILITY

• Got the One-way ANOVA pretty much done
• Went to the Site to get access to the server and discovered VISIBILITY server was down. Tried to send an email to vset but it bounced. Ended up sending an email to John W. Pick that up tomorrow after class
• Sent an email to Denise about getting root access to the new VISIBILITY server
• Started working on F* and W ANOVA variants.
• All this statistics got me thinking. What if the amount of instructions it took to transform the two datasets until they were equal was tracked? Something along the lines of applying an n-dimensional convolution filter to n-dimensional histograms…

7:30 – 5:30 VISIBILITY

• Sent Christine some screenshots of my setup for RemoteObjectTestbed
• Server meetings this afternoon – looks like the goal is to have the server up and ready for us to load things on it by the end of April
• Clean up some of the R code and make a correlation widget that accesses a VISIBILITY table.

7:00 – 7:00pm VISIBILITY

• Walked through the new version of PPM with Christie, she thinks it’s fine, and will check with Trish. I’ll follow up at 3:00, when I have my meeting with Kris U.
• Working from home this morning, waiting for my microwave.
• Looks like the reason for the slowness of the RServe response was that the HashMap that I had the RConnection object wasn’t static. Odd, since the remote object appears to be persistent…
• I have a new microwave!
• R really hates windows paths – you have to make sure to replace all the instances of ” with ‘/’ or you kill R.
• Got all the pieces working. Calculations, plots, the works. Need to do some clean up and commenting, and then make widgets!

7:00 –  5:00 VISIBILITY

• Somehow I lost last Friday’s entry.
  • Walked through the PPM app with Trich, Christie and Leah. There were some problems with how the spreadsheet was getting assembled, but they liked the message organizer.
  • Went over the changes with Dong. He finished up on Friday, and we’re testing today
  • Having very slow response from R when the remote object is running in the server. Odd, because response was immediate when the app was running in java directly. Have to poke at that.
• Testing Dong’s update
• Install this morning?
• I may or may not be leaving this morning. I’m getting a range installed, but the delivery of the range may have been canceled.
• Added RFlexServer and RFlexClient to SVN
• Deployed the new PPM but items that had dollar figures did not post while items that did not have dollar figured *did* post. Talked to Dong about this,  and added the bug to the PPM list.
• Got RFlexServer and RFlexClient running on my home box. Started working on why responses are so slow. THey are not, if R is not being called. Going to try making the reference static.
• Got a new version of PPM from Dong and deployed it. It seems to be working, in that spreadsheets seem to have the appropriate numbers in them :-). I’m going to talk to Christine and Trish tomorrow.
• Set up a meeting with Kris U. tomorrow at 3:00

7:30 – 5:30 VISIBILITY

• Had a thought about getting data out of R. In the same way that I’m writing pix out to files that will then  be read.
  • Tried it, but no luck. It appears that the output from RServe is not considered redirectable
• Went looking for more hints and found this in the RServe FAQ:
  • String s=c.eval(“paste(capture.output(print(summary(mymodel))),collapse=’\n’)”).asString();
  • And that seems to work nicely. The moral of the story seems to be “Find a way to get R to do the work”
• Starting on the RemoteObject – should be mostly finished
• Started RFlexClient
• Accessing the setupModel.xml file
• Having some problems getting the message broker. Found a redundant entry in the remoting-config.xml file. That, plus a recompile, redeploy seems to have fixed it.
• Connected to the RemoteObject! Time to make some fancy widgets 🙂

7:30 – 3:30 VISIBILITY (3.4.11)

• Hoping to deploy new PPM today, and that there will be no mysterious 502 error message…
• Starting on some test statistical widgets to try out the remote object methods
• Deploy went well. Nearly there
• The queries are SLOW. Wondering if it has something to do with getting the RConnection. Going to try bouncing back the request on Monday (Depending on the stove delivery…)

7:30 – 4:30 VISIBILITY

• Went over the latest version of PPM with Dong. We have a bit of a problem saving out multiple files without having a dialog pop up for each file. It’s possible that an Applet might be a workaround
• Working on figuring out how to properly traverse a REXP object
  • Attr
    • Parse the debug string?
      • Split on commas
      • Stuff to the left of ‘=’ is an attribute
      • use getAttribute()
      • Recurse to parser?
  • Payload
    • Object o = rl.elementAt(i);
• Attempted to install PPM on site, and had the following bugs:
  • When running on Trish’s machine, I got a remote object error 502 when trying to save a report for the first time. This did not happen on my box, but it has a static IP address and a hole in the firewall just for it. Will try on another machine today
  • When bringing up a report request, the fields were blank until some event (i.e. show all) was clicked
  • XLS files did not contain dollar figures

9:30 – 7:00 VISIBILITY

• Dentist this morning
• Testing Dong’s code. It looks good as far as it goes. We need some kind of excel export for the financial info.
• Added a page for PPM req’s (for 2011) : http://www.fgmdev.com/blogs/?page_id=4987
• Somewhat useful tutorial for RServe: http://www.ci.tuwien.ac.at/Conferences/DSC-2003/Drafts/Urbanek.pdf
• Here’s another: http://www.stat.auckland.ac.nz/~iase/publications/17/C340.pdf
• Building a more robust data.frame reader. Done
  • It turns out lists can be recursive. I temporarily have the DataFrameReader calling itself, which kind of works but could be better. Fix tomorrow.
• Working my way through getting some basic calculations. There are some issues where if a string is assigned it is not considered a factor (whatever that is…). Unlike in the R interpreter, a factor must be explicitly set (i.e. “menuType <- as.factor(menuType)”)
• Went to the customer site and sat around for two hours.

7:30 – 4:30 VISIBILITY

• It’s starting to feel more Spring-like than wintry. Got 107 miles in on the bike this weekend. My legs are very, very tired.
• Going to try to do some compound actions with RServe, including running a query and generating a plot into a particular directory. If that works, I’m going to put that code in a remote object and try to trigger it from Flex.
  • Making progress, but getting data out of Rserve data frames is very goofy
• Cloud-based IDE? – http://www.cloud9ide.com/. Some more stuff here: http://query7.com/cloud9-ide-review
• Worked with Dong on getting the PPM requests organized in a by-month fashion.