Open Bug 1764662 Opened 2 years ago Updated 10 months ago

Starting the browser starts the socket process very early

Categories

(Core :: Networking, defect, P3)

defect

Tracking

()

People

(Reporter: jld, Unassigned)

References

(Blocks 1 open bug)

Details

(Whiteboard: [necko-triaged])

This code in the DocumentChannel constructor will cause the socket process to be launched (and probably other initialization that could maybe be done later or async?) the first time it's used, and my understanding is that that happens fairly early, the first time we try to render any browser chrome. I don't know how significant this is, but it might have implications for perceived startup performance.

I noticed this in a case where Firefox would try to start, notice that the profile was locked, but fail to connect to the existing instance; in that case it just did a chrome alert about the error (and didn't ever display browser UI), and I had a system call tracer attached so I saw the socket process being started, which seemed unexpected in that situation.

Blocks: socket-proc
Severity: -- → S3
Priority: -- → P2
Whiteboard: [necko-triaged]

Maybe we can wait to launch socket process until the first http load happens.

Priority: P2 → P3
You need to log in before you can comment on or make changes to this bug.