Meeting Notes for 6 August 2008
Present: Mihael, Eric, Liz, Tom L.
Absent: Dan K, Mike W., Randy, Kris, Joao, Tom J., Bob
We had some minor problems at the beginning because we did not have a reservation. Tom’s had run out. Eric set it up, but we got a late start.
Since everybody is busy next week there won't be a meeting. Next meeting is 20 August. Eric has made phone reservations through September, with the access codes added to the Calendar
Mihael could not find them.
Progress on CMS with Swift –
- Had to compile ROOT5
- It works somewhat. Code needs to go into Trunk. May have a new branch later.
- Can run OGRE scripts. Still needs to fix it so OGRE can start ROOT.
Eric: LIGO stuff needs to start ROOT. (ROOT4). Eric needs to rebuild ROOT on Teraport. So maybe Eric's work to get ROOT working for LIGO to run on Teraport will also help get CMS working on Teraport? It depends on whether OGRE would work with ROOT4. (Eric would like to update to ROOT5, but right now LIGO stuff all works with ROOT4 and he doesn't want to change it while he's working on it.)
- Had to hack at OGRE. OGRE will be in SVN. Mihael needs to communicate with Dan on this. Eric will put Bluestone in SVN in the future when Bluestone is at v1.0. It’s currently in his CVS at Spy Hill (see Bluestone CVS info). Mihael could have access to Eric’s CVS in the meantime. It's open to public reading, and Eric can give him an account for writing if he wants.
Registration Bug - MH & LQ
Liz has fixed the bug that Tom J. noticed and Mihael diagnosed. She also corrected an issue of the code making the user directories before it checked if the group already existed.
Wiki Skin Work for Mashups
He modified the "body" wiki skin, to completely remove the closing body and header. It only provides the part that would go into a mashup. He likes the idea of caching that. Then there could be a PHP script to refresh the cache copy
Liz: do we really need to cache it because you don’t see the results in the e-Lab until the cache is refreshed.
Eric: exactly. If we want that, we can use the cache. If we don't want that, we can have the e-Lab page get the body content directly from the wiki on demand (ie. a real mash-up), though right now I don't know how to do the second. But in any case, it's fairly flexible, and we can decide how to best use it.
Still having problems on saving additional metadata when uploading Bluestone plots, but he's at least able to set the upload type to plot
, which means that the user can then view it with their list of plots.; he is tabling the saving of the metadata for the plot until he solves more pressing problems. Mihael said the method used in the
is not as up to date as that used in
, so Eric should emulate that. We discussed whether the caching that Mihael does in VDS could be a problem for this, but it probably isn’t because a new file name is used each time you test.
The current method of saving a plot fails miserably on tekoa, the production server, because it has an older version of PHP without the software needed for the server to do http connections. In the newest version of the software it checks for that problem and just declines to save the plot, instead of crashing.
To save a plot you have to give Bluestone your research group name and password, so that it can save the plot to the e-Lab. It seems like you are being asked to log in again, but actually you are just authorizing Bluestone to act as your proxy. Eric wants this to go away. One way to do this is if you are logged in to the BOINC site, where you have an individual log in, then it can save your group name and password to the database, and then later pull it out of the database when needed. That would be more consistent with "single-sign-on", but still work with the current e-lab "group" system.
Gettting data once per hour again, instead of once per day. Yay.
ELabs hire -
Telephone interviews; only two people replied to Mihael (did not CC) They will interview 4.
We need more advertising
Tom Loughran's report
Tom L. did not join us until 11:45 PM, while Liz had to leave shortly after that.
CMS Testbeam e-Lab
He reported in an earlier e-mail that he still needed to put the changes for in Bugzilla with the subject of “Clean up CMS e-Lab”; None so far
Liz will fix them and put the in SVN when he gets them in Bugzilla.
Jeff Shorney and crowd are done with their test and Tom can use the data for his tests. They are also going to make sure that the DAQs have the latest firmware update. Tom J. thinks it is in the shower code and is pursuing it.
LIGO Analysis Enhancement:
Tom L. had an idea for a new transform for LIGO: His students have been getting earthquakes from the USGS site and then looking for them in Bluestone, and Tom thought it would be useful to be able to enter several time intervals at once, and then have Bluestone plot all of those in parallel. Plotting each one individually is somewhat tedious.
Eric had a related idea of having Bluestone scan a channel for evidence of an earthquake, collecting a list of time intervals of interest. The output would just be a list of GPS start/stop times. The current GDS-EARTHQUAKE monitor is supposedly a signal of an earthquake in progress, but it does not seem to work, and Eric does not know why. So students might be challenged to come up with a better algorithm for looking for a peak and then determining whether it's really an earthquake or something else.
The two transformations could initially be developed separately, but then turned into modules which could be linked, the output of one used as input to the others. This won't happen all at once, or anytime soon, but it's the kind of activity we would like to support and develop, and the students would understand how their work on one or the other might fit in to the bigger goal.
-- Main.LizQuigg - 07 Aug 2008
-- Main.EricMyers - 14 Aug 2008