8:00 – 4:30 SR

• Someone stole my last few teabags! Trader Joe’s today…
• Backups – done
• Prep for review tomorrow
  • Get demo-able version of VersionTracker
  • Status report
  • Burn Dong’s system for deployment on integration server
• Spent the entire morning trying to get the DbObjectTestbed server running in Eclipse. It was working just fine when I turned off the machines for the weekend.
  • Committed my code changes and reloaded the project – nothing
  • Did notice that the properties might be corrupted. I could not edit some of the properties tabs.
  • Added JavaUtils as a project that DbObjectTestbed depended on. That forced a rebuild, which placed the .class files I needed in the Tomcat directory.
  • Rebuilt and deployed the VersionTrackerTest Flex project, and after a hiccup (I’m assuming the server was still coming up), everything is now working. Well, sort of. I have to stop the server, rebuild the Java code then restart the server for the classes to get updated in the webapps directory.
• Working on getting a very finicky Date.parse() to work properly so that the milliseconds since the code was compiled to assemble properly.

8:00 – 4:00 SR

• Got all the certs incorporated. I should be able to switch over at any time. Might try that on Friday.
• Helped Chris with his accounts. He needs to change his passwords and attempt to log on to the test and integration servers.
• Work on the xmlparsers today?
  • There are two problems, one with the hash, and one with components that can be turned on and off
  • If the items in the xml are not in the same order, then the hash will differ, which is bad.
  • The different lengths of xml are due to the the fact that the initial xml is not created by interrogating the display item.
  • Dong suggests the solution is to make the id from only the displayName and yField. That won’t guarantee uniqueness, but it’s close. I could add a validator that checks for a unique displayName too.
  • Yay, that seems to have done the trick. For now. Checking in and burning a new disk for tomorrow.
• Did you know that JSON has functions? Check this out: http://www.jayway.com/2013/04/01/three-undocumented-features-of-json-3/

8:00 – 4:00 SR?

• Spent a long time working through keystore and renewing the sever certs. And then I was informed that server certs could not be renewed, contradicting the email and website instructions that say that they can be renewed. Sigh. More tomorrow.
• Got the FlexiChart bug fixed. I wound up adding a Hash class to Utils, and calculated a hash based on the string (with id set to -1). An RSHash is then computed based on the XML string and that result is used to set the Id. That ID hangs around long enough to be useful, although it does get recalculated when the values change, so it’s not a permanent ID.
• Need to add fix from ColumnSeriesConverter to all the other converters.
  • Which also meant adding seriesConfigChanged(seriesConfig) to all the xxxConfig classes and handling that in SeriesConfig.
  • Which also meant handling the initial creation of the id hash in SeriesConfig.addSeries_clickHandler(event:MouseEvent)
  • The initial creation and updates work, but there is a point that the hashes get out of sync. Making the addSeries perform the hash seemed to move the hash downstream one step.
  • Hmmm. Looks like we have a new bug. With an area chart, the item is initially created as
    • <series id=”2387166208″ seriesType=”AreaSeries” displayName=”New Series” form=”segment” fillColor=”0″ fillAlpha=”0.4″ strokeEnabled=”false” strokeColor=”0″ strokeWeight=”3″ strokeAlpha=”1″ yField=”Totals”/>
  • but when “Apply” is clicked, it changes to
    • <series id=”1515104256″ displayName=”New Series” seriesType=”AreaSeries” yField=”Totals” form=”segment” fillColor=”0″ fillAlpha=”0.3″ strokeEnabled=”false”/>
  • Which, naturally enough, leads to a different hash.
  • However, once this makes a “round trip(?)” to TrendWidget, it takes the shorter form and everything works.
• Need to clean up a lot of debugging statements…