Re: Do we track releases or development versions [Was: r-bioc-cummerbund ... upstream/1.1.3-17-gae72ebb]
On Tue, May 8, 2012 at 7:10 AM, Andreas Tille <andreas@an3as.eu> wrote:
> Hi,
>
> when I was dealing with a watch file of a BioConductor package it turned
> out by accident that I prepared it for development releases. Thanks to
> Charles this was fixed and we are now tracking stable releases. Today I
> stumbled upon r-bioc-cummerbund which in the watchfile explicitly stated
> to track development release. I do not consider this a good idea for
> two reasons:
>
> 1. Upstream has its reasons to release stable versions. If there is
> a version that is declared stable we should by default pick this one
> except there are really strong reasons which should be documented.
> (For sure this does not concern prereleases which might be interesting
> before there is any stable release).
>
> 2. We finally do not actually manage to package each and every
> development release. In this actual example of r-bioc-cummerbund
> we are lagging behind *several* development releases - so it is no
> argument to gain for the latest and greatest - we just do not have
> the manpower to even release the stable release (were we are also
> lagging behind).
>
> Could we agree to this general rule? Carlos, any reason for explicitly
> tracking devel? Would you mind updating the package?
>
I agree on following stable releases as a general rule. The reason I
chose to use development version of cummeRbund was because at the time
upstream was recommending it. The package was on heavy development and
stable releases tied to Bioconductor were getting behind quite fast. I
just checked the upstream website and now the recommendation is to use
the stable version tied to Bioconductor.
I just finish my semester yesterday and I plan to get up to speed with
several TODOs I have with the packages I'm contributing to.
I'll do this update right away.
Thanks for the heads up,
Carlos
Reply to: