[Rd] SUGGESTION: Use JIRA for Bug Reporting, Package Development and Project Management

[Duncan Murdoch]

On 08/06/2017 6:31 AM, TELLERIA RUIZ DE AGUIRRE, JUAN wrote:
> Dear R Developers,
>
> I started programming in R just last January, for which I read (, summarized and memorized) 2 incredible R Programming books: "R Cookbook" and "R Graphics Cookbook".
>
> However, I did before know how to program in MariaDB SQL language, and had submitted for their bug platform previously some bugs (https://jira.mariadb.org/projects/MDEV/issues), which is based on JIRA:
>
> https://www.atlassian.com/software/jira
>
> On the other hand, R has the mailing list, which is really simple, but:
> a) I found it really difficult at first to understand how it worked.
> b) I feel it does not fully respect the privacy of the package maintainers.
> c) And it can give place to e-mail Spam.
>
> In fact, Rstudio has disabled bug.report(package = "somePkg") command in order to avoid misuse.
>
> That is why I would suggest that, being JIRA free for Open Source Projects such as R:
>
> https://de.atlassian.com/software/views/open-source-license-request
>
> It would be really worth it to start using this modern platform gradually instead of Bugzilla, or the mailing lists, for new R developed packages.
>
> This would imply a considerable change to the R community, but I really think it would be worth it, and would help it to improve and be even greater. Although I might be wrong, and there might be different point of views which could be better than mine. However, I do sincerely think that testing this platform instead of Bugzilla would be really worth it.
>
> Hope my suggestion is useful.

You are talking about R and its packages as though they are all part of 
one project, but in fact most packages are separate projects.  Many of 
them use other systems for bug reporting and contributions.  Github is 
quite popular these days; many older packages use R-forge, and of course 
Bioconductor provides its own facilities.

I think it's unlikely that R itself would move to JIRA, for the reasons 
Sahil gave, and other reasons (including inertia).  But if you are 
arguing that individual packages should move to JIRA, you need to 
explain why it is better than Github, R-forge, Bioconductor, etc, not 
just better than the mailing lists.  I think it will be hard to make the 
argument, because all of those have R-specific features, as well as 
existing communities to go to for help.

Duncan Murdoch