On Mon, May 06, 2019 at 08:30:40PM +0100, Ken Moffat via blfs-dev wrote:
> On Mon, May 06, 2019 at 12:40:46AM -0500, Bruce Dubbs via blfs-dev wrote:
> > On 5/5/19 10:47 PM, Ken Moffat via blfs-dev wrote:
> > > Trying to build libreoffice-6.2.2.2 I'm getting a load of
> > > /usr/include/c++/8.3.0/bits/ (various files / positions)
> > > error: template with C linkage messages.
> > >
> > > AFAICS, these seems to be a more detailed message with 8.3.0 (i.e. it
> > > points to where the problem occurs, according to the release notes),
> > > but I built 6.2.1.2 with 8.3.0 without problems, and the only
> > > obvious changes since then are that I've used system gpgmepp because
> > > of gawk-5.0, and boost is now 1_70_0.
> >
> > I built libreoffice 23 Apr 2019 and boost_1_70_0 on 18 Apr 2019.
> > I have not updated gawk on my development system yet.
> >
> > I'll try a build with gcc9 tomorrow.
> >
> > -- Bruce
> Thanks. I've so far confirmed that both the previous and current
> versions of LO have the failure (I'm using LFS from 20190420,
> r11583, and BLFS r21496 from two days later, except for current
> icewm and exiv2) - those details are to try to help narrow down
> *what* caused the breakage, and in case you don't have the breakage
> on your newer system ;)
>
> Started pulling out --with-system options for
> things, one at a time. No joy on that so far, but I've now looked
> at the *last* message in the errors, instead of where it started,
> and it looks like one of the external packages is being broken by
> something:
>
> In file included from
> /scratch/working/libreoffice-6.2.2.2/xmlsecurity/inc/xmlsec-wrapper.h:32,
> from
> /scratch/working/libreoffice-6.2.2.2/xmlsecurity/source/xmlsec/errorcallback.cxx:22:
> /scratch/working/libreoffice-6.2.2.2/workdir/UnpackedTarball/xmlsec/include/xmlsec/base64.h:15:1:
> note: 'extern "C"' linkage started here
> extern "C" {
>
> That is https://dev-www.libreoffice.org/src/xmlsec1-1.2.27.tar.gz
>
> According to debian, xmlsec is "Apache XML Security For Java" which
> is particularly annoying because my own builds use '--without-java'.
> NB xmlsec1, not the current xmlsec2.
>
> Looks like https://www.aleksey.com/xmlsec/ has a 1.2.28 release from
> April. It claims to need: LibXML, LibXSLT, and one of OpenSSL or
> GnuTLS or libgcrypt or NSS. The download page notes that 'make
> check' requires an internet connection, and implies it should be
> installed before running the tests.
>
> Will give it a go, but I have more urgent priorities in the next
> couple of hours - for such a small tarball, it's a very slow
> download. But having got it, the ChangeLog points to
> https://github.com/lsh123/xmlsec/commits/master so github might have
> a faster download.
>
> On an initial look, nss also requires nspr, gnutls requires gcrypt
> which I have not installed - seems it will build for everything it
> can. A good name mentioned in the 'HACKING' file, and it suggests
> the conventional make ; make check.
>
> Almost all the tests were skipped. Looks like it will need
> --disable-static. I'll come back to this later : just because it
> looks as if it will build, doesn't mean the problem is not still
> there.
>
The old "hit send, start to think about it" situation:
http://document-foundation-mail-archive.969070.n3.nabble.com/Build-failure-with-latest-ICU-64-1-td4259322.html#a4259359
Will try disabling system icu. Oh, wait - the link to the ICU bug
https://github.com/unicode-org/icu/pull/572 points to
https://github.com/lsh123/xmlsec/commit/0a14f52e9a5976c104d0fc2d1f739508042aa752
and those fixes appear to be in xmlsec1-1.2.28.
ĸen
--
Before the universe began, there was a sound. It went: "One, two, ONE,
two, three, four" [...] The cataclysmic power chord that followed was
the creation of time and space and matter and it does Not Fade Away.
- wiki.lspace.org/mediawiki/Music_With_Rocks_In
--
http://lists.linuxfromscratch.org/listinfo/blfs-dev
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
