[Debian-med-packaging] Bug#927166: Any chance to get libbiod compiling again?

[Andreas Tille]

Hi,

[Pjotr as upstream in CC]

On Wed, Sep 18, 2019 at 04:54:16PM +0200, Matthias Klumpp wrote:
> Am Fr., 13. Sept. 2019 um 22:51 Uhr schrieb Andreas Tille <andreas at an3as.eu>:
> >
> > On Fri, Sep 13, 2019 at 03:37:32PM +0200, Matthias Klumpp wrote:
> > > Am Fr., 13. Sept. 2019 um 15:21 Uhr schrieb Andreas Tille <andreas at an3as.eu>:
> > > > /usr/bin/ld.gold: error: bin/biod_tests.o: multiple definition of '_d_execBssBegAddr'
> > > > /usr/bin/ld.gold: contrib/undead/*/__main.o: previous definition here
> > > > /usr/bin/ld.gold: error: bin/biod_tests.o: multiple definition of '_d_execBssEndAddr'
> > > > /usr/bin/ld.gold: contrib/undead/*/__main.o: previous definition here
> > > > collect2: error: ld returned 1 exit status
> > > > [...]
> > >
> > > It's very likely that undeaD needs to be updated first in order to
> > > compile this again.
> > > Has upgrading that been attempted yet?
> >
> > I just upgraded libundead and re-added it to libbiod (I assumed libundead
> > would not have been needed for libbiod any more and thus it was removed)
> > Now the error is:
> 
> When I try to compile libbiod now, I get this error:
> ```
> ../bio/bam/baseinfo.d-mixin-165(165): Error: Options at
> ../bio/bam/baseinfo.d(579) conflicts with Options at
> ../bio/bam/baseinfo.d(482)
> ../bio/bam/baseinfo.d(175):        called from here: getRangeMethods()
> ../bio/bam/baseinfo.d(278): Error: template instance
> `bio.bam.baseinfo.PerBaseInfo!(BamRead, "FZ", "MD", cast(Option)0,
> cast(Option)1, cast(Option)2, cast(Option)3)` error instantiating
> ../test/unittests.d(382):        instantiated from here:
> basesWith!(BamRead, MixinArg!(string, "flowOrder"), MixinArg!(string,
> "keySequence"))
> [100/177] ldc2 -I=biod_test at exe -I=. -I=.. -I/usr/include/d
> -enable-color -O -g -release -wi -unittest
> -of='biod_test at exe/bio_bam_cigar.d.o' -c ../bio/bam/cigar.d
> [101/177] ldc2 -I=biod_test at exe -I=. -I=.. -I/usr/include/d
> -enable-color -O -g -release -wi -unittest
> -of='biod_test at exe/bio_bam_tagvalue.d.o' -c ../bio/bam/tagvalue.d
> [102/177] ldc2 -I=biod_test at exe -I=. -I=.. -I/usr/include/d
> -enable-color -O -g -release -wi -unittest
> -of='biod_test at exe/bio_bam_bai_indexing.d.o' -c
> ../bio/bam/bai/indexing.d
> [103/177] ldc2 -I=biod_test at exe -I=. -I=.. -I/usr/include/d
> -enable-color -O -g -release -wi -unittest
> -of='biod_test at exe/bio_bam_iontorrent_flowcall.d.o' -c
> ../bio/bam/iontorrent/flowcall.d
> [104/177] ldc2 -I=biod_test at exe -I=. -I=.. -I/usr/include/d
> -enable-color -O -g -release -wi -unittest
> -of='biod_test at exe/bio_bam_thirdparty_msgpack.d.o' -c
> ../bio/bam/thirdparty/msgpack.d
> ../bio/bam/thirdparty/msgpack.d(2035): Deprecation: integral promotion
> not done for -header, use '-preview=intpromote' switch or
> -cast(int)(header)
> ../bio/bam/thirdparty/msgpack.d(2035): Deprecation: integral promotion
> not done for -header, use '-preview=intpromote' switch or
> -cast(int)(header)
> ../bio/bam/thirdparty/msgpack.d(2035): Deprecation: integral promotion
> not done for -cast(byte)-header, use '-preview=intpromote' switch or
> -cast(int)(cast(byte)-header)
> ../bio/bam/thirdparty/msgpack.d(2035): Deprecation: integral promotion
> not done for -header, use '-preview=intpromote' switch or
> -cast(int)(header)
> ../bio/bam/thirdparty/msgpack.d(2035): Deprecation: integral promotion
> not done for -cast(short)-header, use '-preview=intpromote' switch or
> -cast(int)(cast(short)-header)
> ../bio/bam/thirdparty/msgpack.d(2035): Deprecation: integral promotion
> not done for -header, use '-preview=intpromote' switch or
> -cast(int)(header)
> ninja: build stopped: subcommand failed.
> dh_auto_build: cd obj-x86_64-linux-gnu && LC_ALL=C.UTF-8 ninja -j6 -v
> returned exit code 1
> ```
> This is definitely an error in BioD itself, caused by a deprecation
> that turned into an error in recent LDC versions.
> So, libbiod will need to be upgraded for sure to fix this.

@Pjotr: Can you comment on this?

> In order to support all architectures and generate a pkg-config file I
> would have to reintroduce the Meson build definition (which upstream
> apparently has removed) or apply some hacks around the existing
> Makefile to make that work in a Debian context, which is quite a bit
> more work.

As I told you before:  I have no idea about meson.  It would be great if
we could get it working but if we restrict the package to those
architectures where it builds out of the box and save some manpower I
bet the world will keep on turning round.
 
> Btw, if libundead has no users anymore, removing it completely may be
> a good idea - we don't need to maintain something that's dead and has
> no users.

I was about to file a removal request to ftpmaster before you said in
your last mail that the former build issue might have been caused due
to the lack of libundead.  I would really love to get rid of unneeded
packages.

Kind regards

      Andreas.

-- 
http://fam-tille.de