Skip to content

October 24, 2017, Tuesday

Jump to bottom
Liya Wang edited this page Oct 30, 2017 · 17 revisions
  • Upgrade the local version (master branch)
    • Upgrade to MySQL
    • Upgrade sciapps.org to master branch
      • Done: Remove gzip from maker since we are not archiving any more
      • Done: Remove MaizeCode and SorghumDB from production (Just the type line)
      • Done: Remove Beta apps from agaveAppList.json
      • Done: Change lib/iPC/SciApps.pm line 206: $result=browse_ils($path, $datastore_system, $datastore_homepath);
      • Done: Re-run public workflows, save jsons and update them
      • Done: Revise tutorial in CyVerse Learning Center with PCA based workflow
      • Done: Remove results2 folder (temporarily created for archiving)
      • Done: Not added: Remove VCFvalidator (consider releasing on the Community version)
      • Done: Update manuscript and suppl. for taking the new association workflow
      • Not need since we will distribute the version with SQLite: Update instructions for MySQL
  • Optimizing the community version
    • No response when log in, need to wait for a while, then click on 'Login' multiple times
      • Is the reason that all apps (private and public) are pulling from Agave?
        • Seems it didn't happen again and login is fast
      • If true, we can modify it with following ways:
        • Set up a curated app list to preload like the local version (similarly we can group them by category)
          • Since not all apps are ready for workflows
        • Only load private app when user clicked and add 'more' button so we don't need to load all apps at one time
          • Actually they are loaded behind the scene and it won't block loading of other contents
    • Regarding fixed output filenames
    • Source of data (on the browsing panel)
      • Change 'Browse DataStore' to 'Browse CyVerse Data Store'
      • Remove 'exampleData'
      • Change 'CyVerseUserData' to 'home'
      • Change 'CyVerseSharedData' to 'shared'
      • Change 'archivedJobs' to 'recentJobs'
      • Add page support? Ignore for now
    • Login optimization
      • Can we enable 'return' key'?
    • Other issues
      • Inputs can not be accessed from the diagram no matter its public or not, to be accessible
        • It has to be 'readable' by the world through Agave (do we want to do it? probably not)
        • It can not be a directory
      • User will be logged out after 4 hours
        • Can we renew the token in the backend?
        • Otherwise how do we save a long running workflow?
    • Relax the requirements on fixed output names
      • How about adding output id as prefix? e.g. add 'bwa-index-out' to sorghum.tar.gz as bwa-index-out.sorghum.tar.gz?
        • id is unique in identifying output from the results folder
      • This will resolve the collapsing issue and make workflow more generic, e.g. tools take prefix as parameter
      • Prefix needs to be added to the wrapper script: rename 's/^/bwa-index-out./' *
      • Given bwa-index-out, workflow needs to find bwa-index-out.sorghum.tar.gz and supply it to the next step
      • Challenge is how do we display it? We need to reconstruct the file name! Maybe we can save it in the workflow json
        • Since name can change but output id is fixed
        • Re-run the workflow with different data, we will need to map from bwa-index-out.sorghum.tar.gz to bwa-index-out?
Clone this wiki locally