[Debian-med-packaging] Hmmer2 patch necrothread

[Andreas Tille]

On Sun, Oct 28, 2018 at 10:19:22AM -0400, jrmarsha wrote:
> > 
> >    1. Where did you sent this to?
> I sent it here.  It was rejected upstream.  Upstream was not professional,
> but things still need some work.  It was spurred by a collision with version
> 3 in the Arch User Repository.

Its probably not a good idea to sent a > 2MB patch around via e-Mail.
You should expect that this is rejected in public mailing lists.

> >    2. What's the purpose of this patch
> It drops support for deprecated and no longer supported technologies like
> Parallel Virtual Machine, enabling pthread use by default, associated code
> cleanups, renaming the binaries to what debian-med already does, and
> documentation changes to properly reflect the changes to PVM removal and
> binary renaming.

Sounds interesting.  I admit I'm not a user of hmmer2 - do you think
that its usage statistics are motivating that massive changes.  I wonder
whether it might make sense to add hmmer2 functionality to hmmer3 (may
be that's a naive suggestion - I said I'm not a hmmer user).

> >    3. Could you please strip down the patch to the
> >       technical essence?  If seen lots of documentation changes
> >       and changes of autogenerated files - I have no idea how t
> >       review this.
> There was quite a lot that needed to be reworked.  There is nothing simple
> and complete to really get this patchset right.

$ grep -e '^+++ ' -e '^--- ' hmmer2.patch | less
--- /dev/null
+++ b/.gitignore
--- a/Bugs/BUGTRAX
+++ b/Bugs/BUGTRAX
--- a/CHECKLIST
+++ /dev/null
--- /dev/null
+++ b/COPYRIGHT
--- a/INSTALL
+++ /dev/null
--- a/documentation/userguide/license.tex
+++ b/LICENSE
--- while being no significant hindrance on basic research.
--- a/Licenses/COPYING.celera
+++ /dev/null
--- a/Licenses/COPYING.gnu
+++ b/Licenses/COPYING.gnu
--- a/Licenses/COPYING.pangea
+++ /dev/null
--- a/Licenses/COPYING.spsoft
+++ /dev/null
--- a/Licenses/LICENSE.celera
+++ /dev/null
--- a/Licenses/LICENSE.pangea
...
--- /dev/null
+++ b/Makefile
--- a/Makefile.in
+++ /dev/null

I admit all this looks quite suspicious and intransparent to evaluate
your changes.  If you ask me the way to go would be

   1. Inject hmmer2 sources into some Git repository
   2. Do sensible commits which can be evaluated separately
      (without beeing blured by doc changes, .gitignore files etc.)
   3. Add some release tag

We could point debian/watch file to this and could turn this into a
package.  I'd refrain from blindly adding a >2MB patch without
giving anybody a chance for sensible review.

Thanks for your work on hmmer2 in any case

       Andreas.

-- 
http://fam-tille.de