DBS Discussion Page

we had a meeting with students and post-docs at UCSD to discuss the things people did not like with DBS. Below are the minutes of that meeting.

Further below, I have also added a few things regarding PhEDEx? that irritate me.

As a general comment, people here would love to hear back from you once some of these proposals have been addressed.

Best wishes, Frank

DBS

• Link to DBS

• Marco's comments
  • General and probably not related to DBS: the dataset names are very bad, no convention and random names
    • Would be helpful to alphabetize without respect to capital/lowercase
    • Example of bad naming convenctions
      • DY and DrellYan? and Drell-Yan
      • QCD_Pt vs. QCD_pt
    • People should be forced to fill descriptions when adding data sets
  • Provenence not always possible to follow
  • DBS_Discovery - Navigator
    • I really don't like that with the back of the browser I loose the selections I made
      • Could use cookies to save defaults as well as searches
      • Also, researching within a current search
    • I don't know where to find the cross section/filter efficiency/ (impossible to compute equivalent lumi for given sample)
    • By doing wget of the cff files one gets an html page with embedded cff file and not a text file * DBS_Discovery - Finder
    • Again I really don't like that with the back of the browser I loose what I did before
    • Queries are difficult for me, I did not spend much time but it looks complicated and not very well working

• mnorman's comments
  • The interface needs a category to allow selection by number of events. Sometimes you need samples greater than a validation sample, and sorting through hundreds of samples looking for one that has significant events is difficult.
  • Due to the problems with running at different sites, we probably want a site discriminator again. We need to be able to select either FNAL, or perhaps US sites, or some other differentiation.
    • • Has been found, under "Sites" instead of implemented within "Navigator" or "Finder"
  • Ronin suggested allowing us to search for releases after a certain release.
  • Samples with no data should be removed from the list by default.
  • It should be easier to find either the .cfg file or the generation data for a given dataset. I.e. the provenance of a sample is difficult to figure out because the cfg files aren't available form the browser.
• Johannes would like to be able to select multiple items in each menu (for example, releases 1.6.x)
• Need Cheat Sheet to show us the syntax for AND/OR and RegEx? 's
• Should be able to directly start a Phedex request after finding your data set. Right now you have to follow a bunch of links and then cut and past from one window to the next.
• "Finder" does not seem to work at all - it either finds nothing or everything

Phedex

• Link to Phedex

• None of the defaults make any sense. PhEDEx? knows my DN once I log in. Based on that DN it should set defaults, or allow me to set defaults that make sense to me. Examples are:
  • The site(s) that get shown.
  • The requests for data movement that I have the privilege to sign off on.
• There is one default that nobody ever wants. That's the show all samples instead of showing only the ones that matter.
• At present, there's a huge amount of "spam" hitting the data managers because of subscriptions involving more than one site. When any of the sites do as they are told, all sites get to see the email acknowledgement. I want only email that involves me.

-- MatthewLeBourgeois - 14 Feb 2008