Please see my in-line comments. ----- Original Message ----- > From: "Justin Lebar" <[email protected]> > To: "Jonas Sicking" <[email protected]> > Cc: "dev-webapi" <[email protected]>, "William Chen" > <[email protected]>, "Anne van Kesteren" > <[email protected]>, "Doug Turner" <[email protected]>, "Mounir > Lamouri" <[email protected]>, "EDUARDO > FULLEA CARRERA" <[email protected]>, [email protected] > Sent: Tuesday, May 7, 2013 9:23:03 AM > Subject: Re: Improving system messages to support webpages and more > > > I agree that we need to add a commit() function. A few related > > questions: > > Just to be clear, commit() alone does not solve the question of > idempotence; I don't want us to gloss over that. We have to modify > all of the messages (or at least their semantics) so they're safe to > re-deliver.
We've already had an acknowledgement mechanism very similar to the role of commit(). Please see my previous comment in the same thread and let me know if they're actually different concepts. > > > * Should we up an apps OOM priority until it has called commit()? > > At > > least above other background apps. > > Right now, we give apps higher OOM priority when running their system > message handlers. An app can then request higher OOM priority after > that with a wake lock. I think this is probably fine for now. No matter it's done by a explicit call commit() or the automatic acknowledgement internally done within the SystemMessageManager.js. I'm wondering what is the main concept of this kind of commit()? Does it imply (1) the content process has received the system message or (2) the content process has finished handling the system message? There could be a long time period between (1) and (2) depending on the content design. Gene _______________________________________________ dev-b2g mailing list [email protected] https://lists.mozilla.org/listinfo/dev-b2g
