1191987 - Send iOS formatted Build ID to crash-stats instead of timestamp

Status: RESOLVED FIXED

(Firefox for iOS :: Build & Test, defect)

Description

[Stephan Leroux [:sleroux]]

It was suggested by the socorro team that FennecAndroid uses a timestamp as it's buildid when submitting crash reports and I originally implemented a solution that did that but it turned out to involve some hacks around temporary file creation that results in warnings all over the project when generating builds. It also doesn't get regenerated one every build unless you clean the build folder because Xcode caches the .xcconfig files. Additionally, looking at crash reports on crash-stats is confusing when trying to determine which 'build' a crash is coming from since it uses the timestamp and not the actual build number we use for testflight/aurora. I suggest we use the build number we use for our iOS app releases as the same build id we send to crash stats to make things more sane. The only complication in doing so is the build number would need to be configured from a .xcconfig file instead of a .plist file since the URL we send to crash-stats is configured there and requires the build id as a URL parameter.

Comment 1

[Stephan Leroux [:sleroux]]

Attachment: PR https://github.com/mozilla/firefox-ios/pull/878

Comment 2

[Stefan Arentz | :st3fan | ⏰ EST | he/him]

How about getting rid of the set-build-version script and the step to call it in the documented "Making a Build" process.

Then replace that with a step to modify and commit Client/Configuration/BaseConfig.xcconfig instead?

I think when we make a build, that file should simply always contain the right and current MOZ_VERSION and MOZ_BUILD_ID for that specific build.

Comment 3

[Stephan Leroux [:sleroux]]

I think that would be better too since the BaseConfig file is the only place we should be touching for overall app configuration anyways. The nice thing about the approach is that we don't need to modify the build numbers for each of the extensions since they just use the same configuration value. Would it make sense to amend this patch with the removal of that script and an associated update to the wiki?

Comment 4

[Stefan Arentz | :st3fan | ⏰ EST | he/him]

Yes please.

Comment 5

[Stefan Arentz | :st3fan | ⏰ EST | he/him]

Comment on attachment 8644551 [details] [review]
PR https://github.com/mozilla/firefox-ios/pull/878

THis looks good. Lets do what you suggest in https://bugzilla.mozilla.org/show_bug.cgi?id=1191987#c3 and let the team know that the build procedure has changed a little.

Comment 6

[Stephan Leroux [:sleroux]]

Merged