Re: [platform-vcm-dev] Performance concern using Simple Provider Interfa

[Jean-Michel_Lemieux@xxxxxxx]

"Jim D'Anjou" <danjou@xxxxxxxxxx> Sent by: platform-vcm-dev-admin@xxxxxxxxxxx 31/12/2001 07:33 PM Please respond to platform-vcm-dev

To:        platform-vcm-dev@xxxxxxxxxxx         cc:                 Subject:        [platform-vcm-dev] Performance concern using Simple Provider Interface

> I have a mostly-working version of a simple FTP based repository using the
> interfaces provided in org.eclipse.team.examples.core and
> org.eclipse.team.examples.ui (to complement the existing remote file and
> webdav examples that already exist).

[jean-michel] Cool!

> I have observed that is is necessary contact the server to display the
> context menu for any project resource that is managed using this simple
> team interface. This causes a performance delay due to the server access.
> At this point the user may not need to interact with the server. The reason
> is the hasRemote method the abstract ResourceState class is called,
> presumably to determine if the Team->Get or Put menu items should be
> enabled.

[jean-michel] We had anticipated that the query methods (e.g. isDirty, isOutOfDate, and
hasRemote) should return promptly so that they can be called from the UI. Either cache the remote state or
always return true and handle the error condition in the get/put.

> Currently, this implementation has no practical value due to this problem.
> I looked at an alternative of setting a persistent resource property to
> flag any FTP managed local resource (set after a  download) to eliminate
> t he need for a server call.  However, this approach requires a cleanup
> operation to remove these flags when the Project is no longer connected to
> the FTP server. The "Disconnect Project from Team..." action does not
> provide any notification that would allow me to do this. (separately I
> noticed that persistent properties on local resources were not always
> returned when set. That might be an R2 bug...or mine)

[jean-michel] The ITeamNature::deconfigure() method is called the team nature
is removed from a project.

> I would expect that any implementer would encounter this performance
> problem. If the local resource flagging is a reasonable approach then I
> think that could be implemented in the superclass Resource state (along
> with the cleanup operation). Thoughts?

>Jim D'Anjou

_______________________________________________
platform-vcm-dev mailing list
platform-vcm-dev@xxxxxxxxxxx
http://dev.eclipse.org/mailman/listinfo/platform-vcm-dev