788834 - Upstream uses invalid C++ which newer compilers cannot deal with

Status: RESOLVED DUPLICATE of bug 854389

(Firefox OS Graveyard :: GonkIntegration, enhancement, P5)

Description

[Hubert Figuiere [:hub]]

Attachment: include unistd.h

Emulator arm for b2g failed to build on Fedora 17. Because it hasn't been test on a recent libc.

Attaching patch to sdk/emulator sub tree.

Comment 1

[Hubert Figuiere [:hub]]

Attachment: include unistd.h

Looks like I couldn't manage to create the patch the first time. Here it is this time.

Comment 2

[Hubert Figuiere [:hub]]

And the error I get is a missing getopt() and associated symbols.

Comment 4

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

Current sdk revision is two commits behind aosp/ics-plus-aosp and we hard-coded it in the manifest. https://github.com/mozilla-b2g/android-sdk is long deprecated after switching to B2G-Gingerbread to ICS. We'll need works in both manifests/ and sdk/ to land this. Ask mwu for opinions.

Comment 5

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

Current sdk revision is two commits behind aosp/ics-plus-aosp and we hard-coded it in the manifest. https://github.com/mozilla-b2g/android-sdk is long deprecated after switching to B2G-Gingerbread to ICS. We'll need works in both manifests/ and sdk/ to land this. Ask mwu for opinions.

Comment 6

[Michael Wu [:mwu]]

Host GCCs newer than 4.6.3 are not supported on ICS based B2G.

Comment 7

[Gabriele Svelto [:gsvelto] (PTO)]

Since the build instructions on MDN allow proper building on Fedora 17/18 (with the help of a prebuilt GCC 4.4.7) I think we should close this. See:

https://developer.mozilla.org/en-US/docs/Mozilla/Firefox_OS/Firefox_OS_build_prerequisites?redirectlocale=en-US&redirectslug=Firefox_OS%2FFirefox_OS_build_prerequisites#64_bit_install_examples.3A

Comment 8

[Hubert Figuiere [:hub]]

We should close it when the C++ code is C++ conformant. I know it might take a while with upstream, but the problem is real. It is not just "fancy compilers".

Comment 9

[John Ford [:jhford] CET/CEST Berlin Time]

(In reply to Hubert Figuiere [:hub] from comment #8)
> We should close it when the C++ code is C++ conformant. I know it might take
> a while with upstream, but the problem is real. It is not just "fancy
> compilers".

That's valid and correct, but the reality of the situation is that we don't control the code that's broken.  Renaming this bug to match what is keeping it open.

I suspect that the resolution here will be that nothing changes until we're based on JB Gonk.

Comment 10

[Hubert Figuiere [:hub]]

(In reply to John Ford [:jhford] -- please use 'needinfo?' instead of a CC from comment #9)
> (In reply to Hubert Figuiere [:hub] from comment #8)
> > We should close it when the C++ code is C++ conformant. I know it might take
> > a while with upstream, but the problem is real. It is not just "fancy
> > compilers".
> 
> That's valid and correct, but the reality of the situation is that we don't
> control the code that's broken.  Renaming this bug to match what is keeping
> it open.
> 
> I suspect that the resolution here will be that nothing changes until we're
> based on JB Gonk.

I actually have patches in bug 854389 waiting for a r? that actually solve the problem. yes we can address them.

Marking as dupe.