1360253 - Use lazy MigrationUtils getter and stop using Preferences.jsm from AutoMigrate.jsm

Status: RESOLVED WORKSFORME

(Firefox :: Migration, enhancement, P3)

Description

[:Gijs (he/him)]

Per bug 1329712 comment #2:

(In reply to Florian Quèze [:florian] [:flo] from comment #3)
> (In reply to :Gijs from comment #2)
> > I should really just have removed this in bug 1285577, but it got missed,
> > and then I forgot I'd filed this. But AIUI we can now just request this
> > module when we need it (usually only in startup migration, or when we're
> > showing the relevant "did you want to undo this thing" notifications from
> > about:home or about:newtab).
> > 
> > Given that about:home and about:newtab still load this, it's possible
> > there's more of a perf gain to be had by making that code conditional on
> > automigration being enabled and/or lazy-ifying e.g. the MigrationUtils
> > import (which isn't necessary to just check if we're going to show the
> > notification in the common case that we won't show it).
> > 
> > Florian, is that worth pursuing in a separate bug? Don't know if you've
> > already noticed this in startup profiling or no.
> 
> Here is what I have in my existing startup profiles:
> 
> Mac: https://perfht.ml/2pltLPZ Loading AutoMigrate.jsm takes 3ms, and from
> the profile I would be tempted to say removing Preferences.jsm from that
> module (or at least from the _checkIfEnabled method) may be a good idea; but
> 1ms is just one sample, so can't say for sure.
> 
> Windows: https://perfht.ml/2plmtfc Loading AutoMigrate.jsm is visible but
> just 1ms, so not really significant.
> 
> I should have access to a reference device soon, so I'll have more detailed
> profiles.
> 
> If it's easy, lazy-ifying MigrationUtils would still be a good idea: it
> doesn't seem to be expensive in time (all the modules it imports are
> lazy-loaded), but would still save some memory.

Lazy-ifying the getter and stopping use of Preferences.jsm seems straightforward, so let's at least get that bug on file.

Comment 1

[Matthew N. [:MattN]]

I think both of these issues no longer exist.