Skip to main content

Notice: this Wiki will be going read only early in 2024 and edits will no longer be possible. Please see: https://gitlab.eclipse.org/eclipsefdn/helpdesk/-/wikis/Wiki-shutdown-plan for the plan.

Jump to: navigation, search

Maya 4-05-07

Revision as of 15:59, 5 April 2007 by Tiwebb.cisco.com (Talk | contribs) (New page: == Agenda == # Discussion of Maya Architecture == Attendees == Luc Bourlier, ??? Sharanya Doddapaneni, Cisco Jennifer Li, Cisco Wayn Parret, Genuitec Dave Stevenson, IBM Dennis Vau...)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Agenda

  1. Discussion of Maya Architecture

Attendees

Luc Bourlier, ??? Sharanya Doddapaneni, Cisco Jennifer Li, Cisco Wayn Parret, Genuitec Dave Stevenson, IBM Dennis Vaughn, Cisco Elias Volanakis, Innoopract Tim Webb, Cisco Todd Williams, Genuitec (Update if others missed)

Discussed Topics

Topics Discussed:

Questions during the meeting:

  • What language is the native bootstrap written in?
    • Native bootstrap is written in C. It has been tested on Windows and Linux. There are plans to have it working on Solaris and Mac OSX.
  • Can the launcher manage more than one configuration on a single machine?
    • Yes, the launcher can do this. Maya currently creates a shared cache directory, one per system, this contains all the bundles and features needed for profiles. Maya also creates a configuration directory, which is per profile and named after the profile ID.
  • Built in support for old style bundles?
    • Maya supports plug-ins that have plugin.xml, we convert that to manifest.mf, this is done using Eclipse manifest conversion services.
  • Does the update site have a pluggable mechanism, to add other resources or other types of bundles?
    • The update site is still being worked on to get other sources in for gathering meta information for bundles. The design is aimed at being able to add on other resources. The whole structure to this is still not completely in place. Maya provides software dependency resolver which internally utilizes Equinox resolver to perform dependency resolution. Currently, there is a issue with singleton’s and also sometimes wrong results when resolving the dependencies. This needs to be investigated further.
  • Have you submitted any bugs related to the resolver issue?
    • We still have to determine where the issue is. It’s seems to be an edge case.
  • How to customize profile recommendations for a user?
    • Using weights for a profile, for recommending groups
    • Another way to do this, have identification in the directory, which will let the user to be identified and provide the appropriate recommendation.
    • The roles used to determine the profiles are based on interest level, not administrative level?

Maya currently within Cisco uses mailing lists to determine the right profiles for a given user. The mailing lists with in Cisco are structures to reflect both interests and administrative roles. The behavior for choosing profiles can be changed based on the different organization needs.

  • Explicit administrative role creation functionality for profile recommendation?
    • We don’t have any user interface currently available for that. Important point is to have an extension point to let users implement this functionality.
  • Maya in offline mode, how can this be done?
    • Only run the profiles that were previously run
    • Drop shortcuts for previously run profiles
  • How does Maya persist information on client?
    • Simple files, for better maintainability.
    • There is a need for having a better cache directory on Vista (due to UAC problems) and also on Linux (permissions).
  • How do we deal with users custom add-ons to profiles?
    • Use policies to restrict what profiles can be modified. Policies and decision points should be on the server, to provide some control.
  • Can Maya provide multiple cache locations?
    • This would come into picture for scenarios, where the users adds on software to a given profile. Currently Maya doesn’t follow this convention.
  • Distribution of cache across multiple volumes because of the size?
    • Maya currently just collects software, there is no mechanism to cleanup software.
    • One of the points raised was, there is need for richer metadata for software to better determine the stability of software. Is there a way to this currently?
  • One other point raised was, ease of figuring out how end users can add software to Maya?
    • We currently have plans for a MyProfile perspective, which will provide users to add on update sites. More discussions need to happen around this.

Action Items

  • Document maya.exe and mayaw.exe for the bootstrap
  • On the wiki: Deployment scenarios – cache stured... fix typo

Back to the top