Creating a new hub for MF development activities
 

Hi everyone.

In my absence, I learned to embrace Git/distributed version control, and I can no longer see myself using svn in any way in the future. Additionally, SourceForge, the primary host for a lot of software developed around these parts, has now developed into an ad-and-virus-spewing crapware website like so many others -- what a shame.

Anyways, to fix these issues, to encourage more "drive-by" development from casual users, and maybe to show some people the light of DVC, I've been thinking about rehosting a bunch of this software on GitHub (with proper importing from svn to preserve history). I would most likely structure it as several projects curated by a MersenneForum organization, since a lot of the original developers no longer have the time to be primary maintainers. Of course original authors would get full administrative privileges upon request, and perhaps other, more active members here as appropriate.

Does anyone object to this?

George, does Prime95 have a version of subversion history that includes none of the security code that you otherwise haven't published? If this goes forward I would very much like to include the full development history of Prime95, rather than just starting from a recent snapshot.

Any solution that eliminates SourceForge gets our vote!

If you are thinking of including yafu in this effort then you'll have my support. I know nothing about git/DVC but would like to learn.

Although I do support, it will only work if projects move off of sourceforge. In other words they need to be in only one repository.

I'm thinking all the stuff on here, like yafu, msieve, all the stuff I do, and hopefully Prime95. I know nothing about llr/other prime-sieving/testing related software other than it exists and is a thing, and that rogue develops some of them. :smile:

I would be okay with moving OpenPFGW, PRPNet, and other software of mine to git.

[QUOTE=Dubslow;415036]I'm thinking all the stuff on here, like yafu, msieve, all the stuff I do, and hopefully Prime95. I know nothing about llr/other prime-sieving/testing related software other than it exists and is a thing, and that rogue develops some of them. :smile:[/QUOTE]

I, personally, think this is a great idea.

And I don't agree with "rogue" (deepest of respect) that this will /only/ work if projects move off of sourceforge. It would help in the rankings domain, but isn't *required*; each owner of copyright is, by definition, in control.

If I may give you some advice:

When taking on a project, it is important that you are the champion of that effort. Work as hard as you can; be available to answer any and all questions presented. This is equally important introducing a new solution into an enterprise as trying something new in "meat space".

It gets easier over time, but book at least an hour a day initially (and say goodbye to your non-profit earning weekends fixing SPEs people point out).

In my opinion, doing work like this is the most rewarding most people can do.

Other candidates which escaped me at time of OP are factmsieve.p*, mfakt*, *Lucas*, and... most far-fetchedly, maybe even mesenne.ca or gpu72.com, if their authors are willing to share. (Ideally factordb.com, but its author isn't very responsive right now.) ggnfs-lasieve is another good target. We really are quite a disorganized bunch of louts here, with forks and modifications and tweaks and experiments lying around all higgeldy-piggeldy in whatever tarballs scattered across so many disparate posts. :smile:

Mayhaps Prime95/mfakt*/*Lucas*/gpu72/mersenne.ca/etc should go under a separate GIMPS [URL="https://help.github.com/articles/permission-levels-for-an-organization/"]organization[/URL], a sister to the MersenneForum organization?

[QUOTE=Dubslow;415063]Mayhaps Prime95/mfakt*/*Lucas*/gpu72/mersenne.ca/etc should go under a separate GIMPS [URL="https://help.github.com/articles/permission-levels-for-an-organization/"]organization[/URL], a sister to the MersenneForum organization?[/QUOTE]

Don't overreach, grasshopper...

Take on management of one subproject at a time. You might be surprised how much work is involved in each step.

Sounds like a good idea to me too.