258470 - XULRunner shouldn't use firefox migration service

Status: RESOLVED FIXED

(Toolkit :: Startup and Profile System, defect)

Description

[Jan Varga [:janv]]

Comment 1

[Benjamin Smedberg]

Actually, the xulrunner should attempt to do migration, it just shouldn't use
the firefox migration service. Ick, I'm not sure what we want to do about this
until the siamese twins xulrunner and firefox are separated.

Comment 2

[timeless]

please please please please read component descriptions.
COMPONENT RETIRED. Please do not file bugs here.

Comment 3

[Benjamin Smedberg]

Discussed with darin, we're going to add a DoMigration=true/false key to the
API. This will also prevent a restart in the (common) case that the app does not
provide a migration service.

Comment 4

[Darin Fisher]

Attachment: v0 concept patch

Here's a proof of concept patch.  It added two flags to nsXREAppData:

  enableProfileMigrator
  enableExtensionManager

These flags allow an XRE-based app to enable/disable PM and EM.

I've stripped whitespace changes from this patch.

Comment 5

[Darin Fisher]

Attachment: v1 patch

here's a full patch.  i left composer/app/nsComposerApp.cpp alone since it
seems to have bit-rotted anyways.

Comment 6

[Benjamin Smedberg]

Comment on attachment 158246 [details] [diff] [review]
v1 patch

In general, I prefer "true/false" to 0/1 but I'm not gonna be picky about it.

We still need some kind of compatibility.ini check, even if we disable the EM.
If you like, you can push this off to another bug, it won't be noticed for a
while. Actually, we need to expand the compatibility.ini check, because we need
to look at the xulrunner version *and* the app version.

Comment 7

[Darin Fisher]

> In general, I prefer "true/false" to 0/1 but I'm not gonna be picky about it.

Done.  I'll accept arg[0] == '1', 't', or 'T' as true.  Anything else is treated
as false.


> We still need some kind of compatibility.ini check, even if we disable the EM.
> If you like, you can push this off to another bug, it won't be noticed for a
> while. Actually, we need to expand the compatibility.ini check, because we need
> to look at the xulrunner version *and* the app version.

Yeah, we have the problem that EM is the one that actually writes
compatibility.ini today.  I think we may want to arrange for nsAppRunner.cpp to
completely own the I/O on that file.  That way, we can more easily disconnect EM
if not needed.  I'll file another bug.

Comment 8

[Darin Fisher]

fixed-on-trunk

Comment 9

[Christian :Biesinger (don't email me, ping me on IRC)]

> Done.  I'll accept arg[0] == '1', 't', or 'T' as true.  Anything else is treated

hm... maybe "yes" should also be accepted?