libcmis compile failure

classic Classic list List threaded Threaded
5 messages Options
Ivan Timofeev Ivan Timofeev
Reply | Threaded
Open this post in threaded view
|

libcmis compile failure

Hi,

currently some tinderboxes are red. It seems libcmis encounters a bug in
boost<=1.49:
https://svn.boost.org/trac/boost/ticket/6785
"read_json does not compile on GCC 4.7.0 with std=c++11".

Quoting the report:
"It compiles successfully on 4.7.0 with -std=c++98".

So, should we add this flag for libcmis?

Regards,
Ivan
_______________________________________________
LibreOffice mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/libreoffice
Miklos Vajna-4 Miklos Vajna-4
Reply | Threaded
Open this post in threaded view
|

Re: libcmis compile failure

On Wed, Sep 04, 2013 at 06:16:16PM +0400, Ivan Timofeev <[hidden email]> wrote:
> Quoting the report:
> "It compiles successfully on 4.7.0 with -std=c++98".
>
> So, should we add this flag for libcmis?

Hmm, but we bundle boost 1.54. What's the point of using system boost,
but internal libcmis? Wouldn't it make more sense to tweak the tinderbox
to not mix system and internal libraries?

_______________________________________________
LibreOffice mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/libreoffice

signature.asc (205 bytes) Download Attachment
Ivan Timofeev Ivan Timofeev
Reply | Threaded
Open this post in threaded view
|

Re: libcmis compile failure

On 04.09.2013 19:50, Miklos Vajna wrote:
> On Wed, Sep 04, 2013 at 06:16:16PM +0400, Ivan Timofeev
> <[hidden email]> wrote:
>> Quoting the report: "It compiles successfully on 4.7.0 with
>> -std=c++98".
>>
>> So, should we add this flag for libcmis?
>
> Hmm, but we bundle boost 1.54. What's the point of using system
> boost, but internal libcmis?

A wish to build with as much system libraries as possible maybe. :) We
require quite recent libcmis >= 0.4.0.

Also there is:
AC_MSG_ERROR([--with-system-libcmis conflicts with --enable-dbgutil])

Regards,
Ivan
_______________________________________________
LibreOffice mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/libreoffice
Bjoern Michaelsen Bjoern Michaelsen
Reply | Threaded
Open this post in threaded view
|

Re: libcmis compile failure

In reply to this post by Miklos Vajna-4
Hi,

On Wed, Sep 04, 2013 at 05:50:05PM +0200, Miklos Vajna wrote:
> On Wed, Sep 04, 2013 at 06:16:16PM +0400, Ivan Timofeev <[hidden email]> wrote:
> > Quoting the report:
> > "It compiles successfully on 4.7.0 with -std=c++98".
> >
> > So, should we add this flag for libcmis?
>
> Hmm, but we bundle boost 1.54. What's the point of using system boost,
> but internal libcmis? Wouldn't it make more sense to tweak the tinderbox
> to not mix system and internal libraries?

In general, using system libs might incur high amounts of redtape for
LibreOffice:
- using the external lib might incur bugs/regressions that are not encountered
  with the internal version and requires extra testing or bugfixing
- the external lib must be packaged in a recent enough version
- even a packaged version might need additional review and process to be usable
  by LibreOffice, e.g. https://wiki.ubuntu.com/MainInclusionProcess

as such, having the option to fall back on a internal copy is highly desirable.

Best,

Bjoern
_______________________________________________
LibreOffice mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/libreoffice
Miklos Vajna-4 Miklos Vajna-4
Reply | Threaded
Open this post in threaded view
|

Re: libcmis compile failure

In reply to this post by Ivan Timofeev
Hi,

On Wed, Sep 04, 2013 at 08:15:03PM +0400, Ivan Timofeev <[hidden email]> wrote:
> > Hmm, but we bundle boost 1.54. What's the point of using system
> > boost, but internal libcmis?
>
> A wish to build with as much system libraries as possible maybe. :) We
> require quite recent libcmis >= 0.4.0.

I understand the point of building with all system libs from a distro
point of view -- but when you're hacking, if you want to reduce from
scratch build time with that, I guess using ccache is a better option,
the current JSON problem is an example of that.

> Also there is:
> AC_MSG_ERROR([--with-system-libcmis conflicts with --enable-dbgutil])

Sure, that's true for every C++ lib, like cppunit, etc.

Miklos

_______________________________________________
LibreOffice mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/libreoffice

signature.asc (205 bytes) Download Attachment