945711 - [Messages] We lose the 'body' information when migrating

Status: RESOLVED FIXED

(Core :: DOM: Device Interfaces, defect)

Description

[Julien Wajsberg [:julienw]]

+++ This bug was initially created as a clone of Bug #885679 +++

STR:
* flash a reference workload: make reference-workload-light APP=sms
* launch the Messages app

Expected:
* We should see the body of the last message in the threads that have lastMessageType = 'sms'.

Actual:
* we see the string "undefined" instead.

This is probably a regression from bug 885679, so asking 1.3?.

Comment 1

[Julien Wajsberg [:julienw]]

Note that Rik got it with actual messages (not the workload) when upgrading from a master from october to a master from now.

Comment 2

[gbennett]

.:Last Working Build:.
Environmental Variables:
Device: Buri 1.3 mozRIL
BuildID: 20131106040203
Gaia: 3b5fe429f2414f2a9d7241b311b399033bb10612
Gecko: 9ba3faa35c96
Version: 28.0a1
Firmware Version: 20131115

.:First Broken Build:.
Environmental Variables:
Device: Buri 1.3 mozRIL
BuildID: 20131107040200
Gaia: 42bbe26a72e030faf07a6fc297f61a3a8ccda25b
Gecko: 70de5e24d79b
Version: 28.0a1
Firmware Version: 20131115


Here are the gecko changes for that range courtesy of Naoki: http://hg.mozilla.org/mozilla-central/shortlog/4ed1121bf0f4

Most of the changes are related to browser, but there are a few that might be related: bug 928756, bug 928756, and one backout bug 862849.

Comment 3

[Jason Smith [:jsmith]]

Can someone check if this reproduces on 1.1 or 1.2?

Comment 4

[Julien Wajsberg [:julienw]]

Here is the gecko range: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=9ba3faa35c96&tochange=70de5e24d79b

bug 885679 is actually in this range, which confirms my feeling.

Comment 5

[gbennett]

This does not occur on latest 12/06 Buri 1.2 comRIL.

Comment 6

[Gene Lian (I already quit Mozilla)]

After a DB version 13, the schema updating process will run twice for each version...

I/Gecko   ( 1775): gene test: update: currentVersion: 11
I/Gecko   ( 1775): gene test: update: currentVersion: 12
I/Gecko   ( 1775): gene test: update: currentVersion: 13
I/Gecko   ( 1775): gene test: update: currentVersion: 14
I/Gecko   ( 1775): gene test: update: currentVersion: 14
I/Gecko   ( 1775): gene test: update: currentVersion: 15
I/Gecko   ( 1775): gene test: update: currentVersion: 15
I/Gecko   ( 1775): gene test: update: currentVersion: 16
I/Gecko   ( 1775): gene test: update: currentVersion: 16
I/Gecko   ( 1775): gene test: update: currentVersion: 17
I/Gecko   ( 1775): gene test: update: currentVersion: 17
...

So, in upgradeSchema17(...), the following codes will run twice:

  threadRecord.body = threadRecord.subject;
  delete threadRecord.subject;

making |threadRecord.body| becomes undefined.

Comment 7

[Gene Lian (I already quit Mozilla)]

Attachment: Patch

Let's land this fast...

Comment 8

[Gene Lian (I already quit Mozilla)]

Hi Joe, this bug is a regression. We MUST land this for V1.3+ if V1.3 has already got branched.

Comment 9

[Gene Lian (I already quit Mozilla)]

This is a regression caused by Bug 914060 (V1.2 blocker). I'd strongly suggest let's mark this as V1.2 (koi+) although V1.2 is not affected.

Comment 10

[Vicamo Yang [:vicamo][:vyang]]

(In reply to gbennett from comment #5)
> This does not occur on latest 12/06 Buri 1.2 comRIL.

This is Mozilla RIL.  Please don't set the flags if you don't know what you're doing.

b2g26_v1_2 is certainly affected.
http://git.mozilla.org/?p=releases/gecko.git;a=blob;f=dom/mobilemessage/src/gonk/MobileMessageDatabaseService.js;h=f04829cea0d089a06b42f8b7a3aeedeaf4d82aaa;hb=v1.2#l854

Comment 11

[Vicamo Yang [:vicamo][:vyang]]

https://hg.mozilla.org/integration/b2g-inbound/rev/dec3e196ed65

Comment 12

[Julien Wajsberg [:julienw]]

Great! you're the best, guys ;)

Comment 13

[gbennett]

(In reply to Vicamo Yang [:vicamo][:vyang] from comment #10)
> (In reply to gbennett from comment #5)
> > This does not occur on latest 12/06 Buri 1.2 comRIL.
> 
> This is Mozilla RIL.  Please don't set the flags if you don't know what
> you're doing.
> 
> b2g26_v1_2 is certainly affected.
> http://git.mozilla.org/?p=releases/gecko.git;a=blob;f=dom/mobilemessage/src/
> gonk/MobileMessageDatabaseService.js;
> h=f04829cea0d089a06b42f8b7a3aeedeaf4d82aaa;hb=v1.2#l854

For the latest 1.2 builds we flash using comRIL which is why I responded as such. That's a really rude thing to state even if it is true..what have I done wrong?

Comment 14

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/dec3e196ed65

Comment 15

[Gene Lian (I already quit Mozilla)]

Thanks gbennett very much for testing this. I don't think you did anything wrong from the QA's point of view. :)

The branch of status-b2g-v1.2 is indeed not affected by this bug no matter we use MOZ RIL or COM RIL because this regression is related to the DB schema updating. So, I tentatively mark this as V1.3 at comment #8 but strongly suggest we should mark this as koi+ at comment #9, to solve the existing deficiency on V1.2 from the developers' point of view. It might still cause some performance issues or other potential bugs that we've not yet caught because the updating process runs twice.

Comment 16

[Vicamo Yang [:vicamo][:vyang]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/6bb84d0bc170
https://hg.mozilla.org/releases/mozilla-b2g26_v1_2/rev/43d7b300241a