Re: Help needed in C++ / seqan issue
Hi Gert,
On Wed, Apr 15, 2015 at 10:53:27AM +0200, Gert Wollny wrote:
>
> gbp:error: upstream/1r2.3+dfsg is not a valid treeish
>
> I guess you have to do a "git push --tags".
Done. Sorry for the nuisance.
> I also have two dead links in the source tree
>
> mugsyWGA -> mugsy-seqan/projects/library/apps/mugsy/gcc/mugsy
> synchain-mugsy -> chaining/synchain-mugsy
>
> The latter one is properly set when the build starts, the first one not,
> that is at least not before the build fails.
The upstream tarball is a bit in flux. I removed a lot of stuff and
left these dangling links intentionally as a reminder for myself to make
sure that the needed files will be really built. I admit that's a bit
questionable for other people.
> dpkg-buildpackage -uc -us -b
>
> starts and I get all the errors.
>
> The first series of errors is related to the fact that
>
> mugsy-seqan/projects/library/apps/mugsy/rna_alphabet.h
>
> contains what seems to be a copy of
>
> /usr/include/seqan/basic/alphabet_residue.h
>
> with a slightly different naming scheme. Which means one can remove the
> line 36
>
> #include "rna_alphabet.h"
>
> from mugsy.cpp, BUT! I do not know if the underlying code really does
> the same. That's why I didn't upload this change. Does the package
> actually provide tests?
No, it does not. :-(
I was told that these tools are somehow established and remain
unchanged. Most people seem to simply take the prebuild amd64 binary
(which is also the only supported tarball - source comes from svn).
I guess we need to do our own tests - volunteers would be welcome to
provide a test suite.
I admit I somehow regret that I had the following GSoC idea to late:
Provide test suites for all Debian Med packages
I think this makes a great GSoC project - we should remember this for
next year.
As a temporary solution to see if we would get at least any binary I
applied your suggested patch. We might hide the resulting binary (which
might be only a minor part of mugsy) in some experimental staging area
and lead users via docs to this. Otherwise I see no chance to find this
out.
> The remaining errors are primarily related to differences in the bundled
> seqan and the installed one (I used 1.4.1+dfsg-2). I have no idea how
> to fix these.
I also build against seqan-dev 1.4.1+dfsg-2 as this is the current
Debian package.
> Then there are also a few errors that seem to be related to the more
> restrictive handling of C++ with newer compilers.
I guess the code copy of the old seqan will show several C++ problems as
well. So IMHO trying to work with this is similarly fruitless as trying
to port the code to new seqan.
> I happily step in
> again to fix these if someone else was able to correct the seqan version
> related errors.
Cool. Thanks a lot for this offer
Andreas.
--
http://fam-tille.de
Reply to: