Re: q2-* d/watch files updated
On 25.07.21 22:04, Nilesh Patra wrote:
Howdy
On 7/25/21 10:25 PM, Steffen Möller wrote:
Hello,
there is a 2021.4.0 version tagged that should be found by all the
d/watch files but that was not the case for quite a number of the q2-*
packages. I have updated them and gave them all a quick "uscan
--verbose" to test.
Cool, thanks for doing this.
As you might've noticed, a few of these do not build on 32-bit arches due to
missing python3-skbio dep there, so I disabled salsa CI for these
Nice. I mean, not the missing dep, but .. you know what.
Also, there's this package: "q2-vsearch" that you pushed in a commit to, this is
not in the archive -- the build fails too as seen in CI too. Is this useful?
Should this be fixed and uploaded (to NEW)?
I had a look but did not immediately grasp why this failed. In
principle, yes, should be fixed and uploaded. How ultimately useful this
is I cannot tell, really. My microbiome skills are a bit, but not much,
above those of our pets. :-)
@Andreas, https://blends.debian.org/med/tasks/bio does not give the
"Newer update" note for any of these packages. But it should have for
those for which uscan has worked fine, right?
This is because github changed its fetch URL from ./archive/... to ./archive/.*/....
Similar results could be seen on UDD as well[1] -- you can see several
"uscan returned an error: In debian/watch no matching files for watch line https://github.com"; there
I also sent in a mail about it to mass fix watch files[2], and there was a longish discussion on -devel too about this[3]
[1]: https://udd.debian.org/dmd/?debian-med-packaging%40lists.alioth.debian.org#todo
[2]: https://lists.debian.org/debian-med/2021/04/msg00025.html
[3]: https://lists.debian.org/debian-devel/2021/03/msg00179.html
I think it was a mixture of that transition and a copy'n'paste of a
dependency on a github.com/.../release/latest URL that does not seem to
be preserved.
This qiime package and its modules is one of those corners of Debian Med
that keep nagging me. The packages are now all a bit dated, looking
forward to a working unstable and then backports, really. We just need
one regular experiment to run as part of the CI tests and we are good, I
tend to think.
Best,
Steffen
Reply to: