Comment 9 for bug 1385410

Revision history for this message
Seth Arnold (seth-arnold) wrote :

I'm sorry this was overlooked for so long.

> A. Is it practical to use the device while precompilation is happening in the background?

It Depends. We have some options in how aggressively we use system resources, but probably even compiling one-at-a-time with a nice value will definitely feel like the system is being used. Probably using PIM functions will be fine, office productivity might show some lags, games are probably not going to work.

> B. What drawbacks, if any, are there from using the device after precompilation finishes, but before the device restarts?

I believe no drawbacks at all.

> C. When recompilation is required, what proportion of overall update installation time, on average, is taken up by (i) the precompilation (ii) the restart (iii) the flashing (iv) anything else? (Bonus points if you manage to relate (i) to the number or size of the policies.)

I'm sorry to not have numbers to give you, but my recollection of anecdotal evidence suggests it can be fifteen minutes or more. My recollection of reboot times is roughly one minute, and no idea about flashing.

On my core i7 laptop, apparmor_parser takes 2.2 seconds and 20 megabytes of memory to compile the evince profile, typically one of the more expensive desktop and server profiles. I believe it is a fair approximation of a touch/snappy profile.

I hope this helps.