E/GeckoConsole( 76): [JavaScript Error: "[Exception... "Could not convert JavaScript argument arg 2 [nsISettingsServiceLock.set]" nsresult: "0x80570009 (NS_ERROR_XPC_BAD_CONVERT_JS)" location: "JS frame :: jar:file:///system/b2g/omni.ja!/components/RadioInterfaceLayer.js :: setPreferredNetworkType :: line 658" data: no]" {file: "jar:file:///system/b2g/omni.ja!/components/RadioInterfaceLayer.js" line: 658}]

Comment on attachment 663767 [details] [diff] [review] Fixing the error landitlanditlandit Thanks!

Sigh, does *nobody* test their code anymore?

Backed out because we decided to back out bug 787420 altogether: https://hg.mozilla.org/integration/mozilla-inbound/rev/3abe394b6110 Nevertheless, thanks for the patch, Alexandre!

Resolving this bug. We should fold this fix into the new iteration of bug 787420.

https://hg.mozilla.org/mozilla-central/rev/5dc812bc69dc Please don't resolve bugs that haven't landed on m-c yet.

(In reply to Ryan VanderMeulen from comment #6) > Please don't resolve bugs that haven't landed on m-c yet. Yup, normally I wouldn't, but this was a follow up fix for a bug that was backed out. So yeah. :)

(In reply to Philipp von Weitershausen [:philikon] from comment #2) > Sigh, does *nobody* test their code anymore? I apologize for the troubles made. There is always a good reason for one more test case, and here is another. But I did test the patch in my own way, and that's why I attached a "Gaia debug patch" on bug 787420. I verified switching settings in between the three options and rebooting with each of them. All have expected results and then I submitted review requests to you. We can't have marionette test cases for this for now. The reason was already given in the comments. In fact, I did try to write one before submitting review requests, had a brief study on several release versions of Android RIL/QEmu code and the effort to cherry-pick related parts, and found that we just can't achieve it right now. Yes, I missed one more thing, to verify without Gaia patch. Some colleagues and I agreed that we should assume the Gaia doesn't always behave as Gecko might expect the day before. We should not make any assumption on the existence of something, the sequence of events, or the format of, and the relationship between settings values, etc. from Gaia. Still I didn't verify the Gecko patch alone without Gaia part. I guess I have to etch it in the wall.