[Debichem-devel] Updating jmol

[Ximin Luo]

Michael Banck:
> Hi Ximin,
> 
> there's a new upstream release of JMol available:
> https://qa.debian.org/cgi-bin/watch?pkg=jmol
> 
> I had a stab at uscan -vv, but it does loads of scripted stuff that you
> setup long ago, and in the end, I get the current tarball, not the new
> one.
> 
> Can you please update the package and/or quickly explain how it is done?
> 

Hi Michael,

As you noted the script actually totally ignores the version detected by uscan, because this tarball does not contain complete sources for everything - instead it is a binary tarball with some partial sources. For more background please read debian/README.source and https://lists.debian.org/debian-med/2016/11/msg00083.html

Last time I checked the project was in a bad state having accumulated several decades of legacy cruft, and unfortunately fresh academic students were being paid (presumably not very much) to perform the most inane software maintenance tasks on it. I can't imagine the situation has improved much, but perhaps it has.

You should first manually download and review the binary tarball (as detected by uscan) and see the horrific situation for yourself. **Assuming the situation has not improved**, then you can try the following:

1. Edit debian/make-upstream-tarball.sh and change the "Check out source code from SVN" section to point to the correct SVN revisions.

How does one get the correct SVN revisions? For Jmol it should be obvious, for the other components I think I originally did that by downloading the binary tarball, and then reverse-engineering the correct SVN revisions by manually comparing them. It is a time-consuming process but hopefully takes no more than an hour.

2. The other parts of the script may also need to be updated, such as the "rm -rf" parts and the "mv_data" parts.

I tried to get upstream to clean their act together on the sourceforge bug tracker, and they were somewhat responsive, but the conversation was slow and they didn't seem to understand a lot of what I was saying. Perhaps you can try again.

I myself am only interested in Jmol because of Sagemath, which uses it only as a 3D viewer. There is work on Sagemath to move away from Jmol completely, and I hope that will get done in the near future.

X

-- 
GPG: ed25519/56034877E1F87C35
GPG: rsa4096/1318EFAC5FBBDBCE
https://github.com/infinity0/pubkeys.git