[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Upstream concerns about unarr packaging (Was: Newbie questions on packaging opencpn)



Hi Bas, Alec and the rest of the team.

My concern isn't with Alec using my CMake build system,

but with the combination of it's current iteration with the legacy unarr code.

Historically, my fork was intended to just be used in just that way - just add an extra

CMakeLists.txt and you're good to build a shared library.


However, as the original author declared that he didn't have the time to maintain the

code anymore I took the opportunity to further improve things like symbol export, updating

embedded lzma-sdk source code, fixing smaller bugs, add some minor features and so on.

As you already noticed, this broke compatibility with Zeniko's unarr.


Now, what makes me concerned is that instead of investigating the differences between my

fork and unarr classic and doing a proper backport, the CMakeList.txt was modified just enough

to make the code build, apparently without really understanding the mechanisms of the build

system I put in place.


I haven't dug too deep into this, but from what I can tell the package probably ships an incorrect

pkg-config file (unarr classic can't use liblzma, but the build system was modified to both use

the internal lzma code *and* include liblzma as a dependency in pkg-config). There might be other

issues too.


The easiest way out of this would be probably if you just use my updated unarr code for the whole package.

I already provide .deb packages for users[0] and I'd be more than happy to hand over this task to a downstream

maintainer. I'm also still maintaining and improving this code, so I really could use some downstream feedback.


Alternatively I could help you backport the bare necessities to make the CMakeList.txt work cleanly with unarr classic.


Best regards,


Felix


[0]

https://github.com/selmf/unarr-deb

https://software.opensuse.org/download.html?package=libunarr&project=home:selmf


Am 27.09.18 um 21:46 schrieb Sebastiaan Couwenberg:
Hi Felix,

Your concerns from IRC have been noted [0].

Alec is the packager of unarr in Debian and the one that chose to use
your CMake build system in the Debian package.

Can you discuss your concerns with Alec on this list to keep the rest of
the team in the loop?

Kind Regards,

Bas

[0]

14:30 < selmf> hello
14:44 < selmf> anybody here?`:)
14:56 < Myon> selmf: do you have a question?
14:57 < selmf> sort of
14:59 < selmf> there is a package prepared by the GIS team for Debian
which uses some code from me in a way that makes me feel really
uncomfortable
15:01 < selmf> to be precise, this is about
https://salsa.debian.org/debian-gis-team/unarr
15:04 < selmf> an the code in question is a cmake build system I wrote
for my personal fork of unarr
15:09 < selmf> the gis package is currently trying to adapt this build
system to the unmaintained upstream project without taking into account
the changes I did to the actual source code, and this is a bit ugly
15:13 < selmf> so I'm looking for a good way to politely discuss the
issues involved and help getting this on the right track
15:18 < Myon> selmf: the mailing list is probably the better place
15:19 < Myon> or maybe leave a comment/issue with the salsa project
15:31 < selmf> ok



Reply to: