On Mon, Nov 30, 2009 at 11:53 AM, Stefan Kuhn <stefan.kuhn@...> wrote:
> I think there are two issues here:
> 1. Why have two repositories? Well the reason is it was decided by the
> community not to have jcp in cdk.
To complement: the JChemPaint *application* code is not in the CDK
library. The renderer and editing functionality is. The *application*
code is not library code, and, for example, widget set depending.
(application := referring to wrapping GUI code, so including applets,
>> that hasn't been touched in 4 months. That
>> makes this an unlikely candidate for the repo for the most active branch of
>> the project...
> This has been moved to GIT, but JCP is using master now.
I will remove that SVN branch.
>> One final gripey question about the current state of the repos is "why so
>> many rebases for egonw's commits?" rebase is one of those features of git
>> that I try to avoid using. When I see this sort of thing it suggests to me
>> that there was probably a better way to achieve the same ends. But, then
>> again, I don't really understand the relationship between, say, egonw's
>> changes and jcp-primary.
> I can't comment on that, please let Egonw explain this to you.
Rebasing is the git way of automatically updating the patches for
changes for code you depend on. The alternative is to manually rewrite
those patches, which practically can be needed (and often is the
case), when git cannot resolve the differences automatically.
If you have ever took 100+ patches written for tool X version Y and
tried to apply them to version Z (with API changes) of tool X, you'll
start to appreciate rebasing.
>> SUGGESTION 4: Lose the jchempaint-primary name. It's rather confusing to
> This is actually lost.
> I think we should delete the jcp-primary branch from
> svn. I see no reason not to do that. Egonw, what do you think?
Yes, see above. It was kept around for a while until I was 100% sure
we had everything in git, but I forgot to remove that
jchempaint-primary branch in SVN later ...
Regarding the 'jchempaint-primary' name... it has a long history and
should be replaced, and I am open to suggestions. If none show up,
I'll convert it to 'cdk-jchempaint', like the mailing list.
>> I hope this doesn't all come across as too much whiney complaining from a
>> lurker, but I thought it might help to hear how the project(s) look(s) to
>> someone from the outside. I have high hopes for CDK, but I'd rather see
>> developers' energy go into things like proper SMILES parsing and 2-d
>> rendering of geometric isomers, to single out a feature that prevents me
>> from using any version of CDK for real work at the moment, than sorting out
>> the various repo/fork issues.
>> I think it's neither surprising nor necessarily bad that folks want to have
>> different applets/applications that use CDK, even for similar purposes.
>> What would be nice would be to continue factor shared components out of
>> these. If the EBI group wants their own visualization app, great, but let's
>> make sure the underlying foundations are solid.
> I can only share this view. But in order to do that sharing we need a working
> development setup.
We have had two development models for the new JChemPaint architecture
now. One involved SVN which did not work for the Uppsala team, one
which involved Git which does not work for the EBI team.
Post-doc @ Uppsala University