1621024 - Skip reporting errors if the file is not found in the L10nRegistry.loadSync

Status: RESOLVED FIXED

(Core :: Internationalization, defect, P2)

Description

[Jim Mathies [:jimm]]

The payment request dialog currently fails to show on Windows due to a fluent related issue. The failure occurs in the form auto fill initialization here -

https://searchfox.org/mozilla-central/source/browser/extensions/formautofill/FormAutofill.jsm#129

stack:
[JavaScript Error: "[Exception... "Component returned failure code: 0x80520012 (NS_ERROR_FILE_NOT_FOUND) [nsIChannel.open]" nsresult: "0x80520012 (NS_ERROR_FILE_NOT_FOUND)" location: "JS frame :: resource://gre/modules/L10nRegistry.jsm :: L10nRegistry.loadSync :: line 774" data: no]"]
L10nRegistry.loadSync@resource://gre/modules/L10nRegistry.jsm:774:12
fetchFile@resource://gre/modules/L10nRegistry.jsm:676:31
generateResourceSetSync/<@resource://gre/modules/L10nRegistry.jsm:581:19
generateResourceSetSync@resource://gre/modules/L10nRegistry.jsm:576:22
generateResourceSetsForLocaleSync@resource://gre/modules/L10nRegistry.jsm:412:44
generateBundlesSync@resource://gre/modules/L10nRegistry.jsm:174:7
touchNext@resource://gre/modules/Localization.jsm:167:37
onChange@resource://gre/modules/Localization.jsm:500:20
Localization@resource://gre/modules/Localization.jsm:227:10
getRegionDisplayNames@resource://gre/modules/mozIntl.jsm:736:19
@resource://formautofill/FormAutofill.jsm:129:36
get@resource://gre/modules/XPCOMUtils.jsm:129:51
exposeUtilityFunctions@chrome://payments/content/paymentDialogFrameScript.js:96:7
sendToChrome@chrome://payments/content/paymentDialogFrameScript.js:158:12
handleEvent@chrome://payments/content/paymentDialogFrameScript.js:64:10
sendMessageToChrome@resource://payments/paymentRequest.js:76:14
init@resource://payments/paymentRequest.js:28:10
@resource://payments/paymentRequest.js:354:16

Comment 1

[Jim Mathies [:jimm]]

The path to the resource is -
resource://gre/localization/en-US/toolkit/intl/regionNames.ftl

Odd because I can open this in both nightly and my local build.

Comment 2

[Zibi Braniecki [:zbraniecki][:gandalf]]

Understanding regression range would be helpful here.

This error indicates that some file cannot be fetched synchronously from MozIntl, called in the chrome process.
My first guess is that maybe the call is fired too early?

Jim, can you provide STR?

And if you can instrument https://searchfox.org/mozilla-central/source/intl/l10n/L10nRegistry.jsm#758 to add console.log(uri) then we could learn what URI is failing.

As a side note, I'd like to understand why this exception fails to be caught by the try/catch it's in. This seems like a JS engine bug

Comment 3

[Zibi Braniecki [:zbraniecki][:gandalf]]

Oh, thank you, yeah. Ok, so nothing unusual in the fetch, mozIntl requires it.

Is the payment request dialog fired early in the startup or during app's life cycle?

If it's during life-cycle, can you open browser console and call Services.intl.getLocaleDisplayNames(undefined, ["pt-DE"]); before launching the payments dialog?

This should fetch the same file via the same MozIntl.

Comment 4

[Jim Mathies [:jimm]]

Looks like it may have come through this loadSync call twice, failing on the first, succeeding on the second call. I'll keep poking at it.

Comment 5

[Jim Mathies [:jimm]]

Attachment: wierd outline.png

I actually seeing what appears to be a dialog outline painted in chrome. Really odd, it looks like something is displaying but it's broken.

Comment 6

[Matthew N. [:MattN]]

Right, that's bug 1616863

Comment 7

[Zibi Braniecki [:zbraniecki][:gandalf]]

Looks like it may have come through this loadSync call twice, failing on the first, succeeding on the second call. I'll keep poking at it.

Does it mean that the locale shows up, just with an error in the console?

The logic makes sense - we first try to load the file from browser's omni.ja, and only then from toolkit's omni.ja.
The only question is why the exception is not being caught by the try/catch then.

Comment 8

[Lina Cambridge [:lina] (ex-Mozilla)]

The only question is why the exception is not being caught by the try/catch then.

Sorry for the drive-by comment, but I was curious—are you sure it's not being caught? Logging Cu.reportError(Components.Exception('Oops!', Cr.NS_ERROR_UNEXPECTED)) in the browser console produces something similar to that, stack trace and all. I wonder, could it be we're actually catching it, and it's just the formatting from the Cu.reportError call suggesting it's not being caught?

Comment 9

[Zibi Braniecki [:zbraniecki][:gandalf]]

Yep, that seems it.

I guess since sync loading there is fallible, we should just not report the error if the file is missing, since it is an expected scenario.

I'll provide a patch.

Comment 10

[Release mgmt bot [:sylvestre / :calixte / :marco for bugbug]]

The product::component has been changed since the backlog priority was decided, so we're resetting it.
For more information, please visit auto_nag documentation.

Comment 11

[Zibi Braniecki [:zbraniecki][:gandalf]]

Attachment: Bug 1621024 - Skip reporting errors if the file is not found in the L10nRegistry.loadSync. r?mossop

Comment 12

[Bodea Daniel [:danibodea]]

This bug does not have steps to reproduce so a regression range could not be attempted, let alone provide it, but the bug's description shows the code where the request fails and it appears that the bug is already being addressed. This being said, I will remove the "regressionwindow-wanted" tag.

Comment 13

[Jim Mathies [:jimm]]

Updating the title here to reflect the change. Pretty sure the dialog display issue is a dupe of bug 1616863.

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/f29caeedf51e
Skip reporting errors if the file is not found in the L10nRegistry.loadSync. r=mossop

Comment 15

[Arthur Iakab [arthur_iakab]]

https://hg.mozilla.org/mozilla-central/rev/f29caeedf51e

Comment 17

[Release mgmt bot [:sylvestre / :calixte / :marco for bugbug]]

The patch landed in nightly and beta is affected.
:zbraniecki, is this bug important enough to require an uplift?
If not please set status_beta to wontfix.

For more information, please visit auto_nag documentation.

Comment 18

[Zibi Braniecki [:zbraniecki][:gandalf]]

I don't think we want to uplift.