1429345 - unified bustage in bindgen code for PushMessageData

Status: RESOLVED INVALID

(Core :: DOM: Core & HTML, defect)

Description

[Cameron McCormack (:heycam)]

In a try push (that added a new Web IDL interface, and so would have shuffled things around), I got this build bustage on windows-mingw32-32 debug:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=9460241471c313e2309f0504902acc034b61d5ec&selectedJob=155283356

/builds/worker/workspace/build/src/obj-firefox/dom/bindings/PushMessageDataBinding.cpp:275:24: error: 'mozilla::dom::DOMPrefs' has not been declared

PushMessageData has a headerFile specified in Bindings.conf, which means we don't generate a #include "mozilla/dom/DOMPrefs.h" automatically for the [Func] that's used on it, on the assumption that the specified header will manage that for us.  And indeed, ServiceWorkerEvents.h (the headerFile) does #include "DOMPrefs.h".

I'm confused why that's not enough, and setting up a mingw build environment doesn't sound like fun.  Any clues?

Comment 1

[Boris Zbarsky [:bzbarsky]]

> And indeed, ServiceWorkerEvents.h (the headerFile) does #include "DOMPrefs.h".

That was added in https://hg.mozilla.org/mozilla-central/rev/a2b1db70bc89

Your try push is based on https://hg.mozilla.org/try/rev/e4de69553e3f which is not a descendant of a2b1db70bc89 (their common ancestor is https://hg.mozilla.org/try/rev/dc271c111e3352f23b424f0a041432dac027ee78).

I suspect you pulled from m-c after most of the bug 1419771 changesets got merged there, but before the bustage fix in bug 1419771 comment 48 was checked in...

Comment 2

[Cameron McCormack (:heycam)]

Huh, I thought I saw that #include in my local tree.  Maybe I was looking in the wrong checkout.  Anyway, thanks!