Closed rekado closed 8 years ago
I think I fixed it. If it's not fixed, let me know. I'm leaving this issue open until I'm sure it's fixed. Thank you.
It's not fixed.
Look inside the source code release package: https://github.com/smithlabcode/preseq/archive/v2.0.2.tar.gz
You'll see that the smithlab_cpp folder there is empty. I don't think you can expect GitHub to recurse into submodules and fetch the code from them when it's generating the release package.
Yes, we know that GitHub doesn't include submodules on the release. That's why I included a separate tarball (https://github.com/smithlabcode/preseq/releases/download/v2.0.2/preseq_v2.0.2.tar.bz2) that includes the submodule.
The latest release tarball is not self-contained. While it is good that
preseq
does not bundlesmithlab_cpp
, it is not easy to buildpreseq
as there has not been an official release of thesmithlab_cpp
library.Just unpacking the tarball and building preseq results in an error:
In my latest version of the preseq package for GNU Guix I'm first building a prelease of
smithlab_cpp
and then setSMITHLAB_CPP
to the lib output of that package; I also modify theINCLUDEDIRS
variable such that the installed headers of thesmithlab_cpp
package can be found.If there will be a release for
smithlab_cpp
soon, it would be nicer if preseq could be linked against that library directly.