You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently users select which graph to generate by issuing a command line option, to make it more intuitive we could organise the options into "views", such as "Compartments-view", "Threads-view", "Heaps-view", etc. Users can first go to a view, and then interactively generate graphs based on information they want to get from the views.
Perhaps we could have a web interface where view options can be selected, and when a view is displayed, users can zoom in and out of the view or click on the nodes to drill down on that specific compartment or thread, etc.
"I have been pondering visualisations a bit, and I realised that it might be that we want to think about “views” of some sort -- e.g., “compartment centric”, “thread centric”, (perhaps) heap-centric, and so on. In a compartment-centric view we’re interested in the rights accessible from (and to?) thread stacks, but not so much that we’d arrange the whole layout around it. Whereas in a thread-centric view, I think we’d try and lay out all the disjoint stack bits associated with a single thread in vertical “stacks” perhaps with a thread ID / name above, etc.
This might help us focus the ambitions of particular plots a bit more, making it easier to figure out how to arrange them, what to elide/combine/etc."
The text was updated successfully, but these errors were encountered:
Currently users select which graph to generate by issuing a command line option, to make it more intuitive we could organise the options into "views", such as "Compartments-view", "Threads-view", "Heaps-view", etc. Users can first go to a view, and then interactively generate graphs based on information they want to get from the views.
Perhaps we could have a web interface where view options can be selected, and when a view is displayed, users can zoom in and out of the view or click on the nodes to drill down on that specific compartment or thread, etc.
Copied from @rwatson's suggestion on slack:
"I have been pondering visualisations a bit, and I realised that it might be that we want to think about “views” of some sort -- e.g., “compartment centric”, “thread centric”, (perhaps) heap-centric, and so on. In a compartment-centric view we’re interested in the rights accessible from (and to?) thread stacks, but not so much that we’d arrange the whole layout around it. Whereas in a thread-centric view, I think we’d try and lay out all the disjoint stack bits associated with a single thread in vertical “stacks” perhaps with a thread ID / name above, etc.
This might help us focus the ambitions of particular plots a bit more, making it easier to figure out how to arrange them, what to elide/combine/etc."
The text was updated successfully, but these errors were encountered: