Hugin build fails in fedora 30

[tduell]

Building hugin snapshot 8199:62b0662d5bee in Fedora 30 produces the following error...

"/home/terry/rpmbuild/BUILD/hugin-2019.1.0/src/hugin_base/panodata/SrcPanoImage.cpp:339:30: error: expected '{' before '&' token
make[2]: *** [src/hugin_base/CMakeFiles/huginbase.dir/build.make:651: src/hugin_base/CMakeFiles/huginbase.dir/panodata/SrcPanoImage.cpp.o]
Error 1"

The same source code built without error in Fedora 29.
I believe Fedora 30 uses GCC 9.1

[tmodes (tmodes)]

I don't see anything string in the mentioned line. What exiv2 version (released or git version) do you use?

[tmodes (tmodes)]

Typo, I meant: I don't see anything strange in the mentioned line.

[Bruno Postle (brunopostle)]

It appears to be because fedora exiv2 went from 0.27.0 to 0.27.1 a few days ago.

gcc went from 9.0.1 to 9.1.1 a few days before, and this was ok: https://apps.fedoraproject.org/koschei/package/hugin?collection=f30

I had a look at the exiv2 changelog and didn't see anything obvious.

Nothing is significantly different in the build log, the cmake output and compiler flags are the same as before.

[tduell (tduell-iinet)]

I'm using exiv2 0.27.1, as provided by Fedora 30.

[tduell (tduell-iinet)]

Hello Bruno,

On Wed, 15 May 2019 07:53:37 +1000, Bruno Postle
<email address hidden> wrote:

> It appears to be because fedora exiv2 went from 0.27.0 to 0.27.1 a few
> days ago.
>
> gcc went from 9.0.1 to 9.1.1 a few days before, and this was ok:
> https://apps.fedoraproject.org/koschei/package/hugin?collection=f30
>
> I had a look at the exiv2 changelog and didn't see anything obvious.
>
> Nothing is significantly different in the build log, the cmake output
> and compiler flags are the same as before.
>

Are you seeing the same build error?

Cheers,
--
Regards,
Terry Duell

[Bruno Postle (brunopostle)]

On 15 May 2019 00:08:29 BST, tduell wrote:
>Hello Bruno,

>Are you seeing the same build error?

Hi Terry, I haven't had a chance to test myself, but fedora rebuilds every package every few days to check for this kind of thing, so I got an automatic email telling me it was broken - this is how I know that the problem is the exiv2 update not gcc.

[mailing directly because I'm at work and can't remember my launchpad login]

--
Bruno

[tmodes (tmodes)]

I think exiv2 0.27.1 changed the order of the inclusion of the individual files.
So I switch from the inclusion of individual files from exiv2 to the inclusion of the main exiv2.hpp - as recommend in the exiv2 doc.
I hope this fixes the compilation issue with exiv2 0.27.1.

[tduell (tduell-iinet)]

I'm pretty sure that the build now progresses past the reported fail point OK, but still fails to complete OK as a result of python plugin issues.
AS this appears to be unrelated to this bug bug report, I'll raise a new one.

Cheers,
Terry