[gmx-developers] Hidden modules in gmx?

[Erik Lindahl]

Hi Massimo,

We haven't decided anything yet, but the ideal solution (again, IMHO) would be to have a contrib-module that could be downloaded or enabled at compile time, and then you would get a number of extensions that are less tested.

The problem is that while some users realize they need to check their results carefully, other's won't, and then they sometimes end up giving Gromacs a bad reputation for being "buggy", just because they were able to use some esoteric  feature. I don't really have any strong preference how things are set apart, but it is important that we discriminate better between code that _has_ been extensively tested and some random contributed function that might - or might not - work.

At the end of the day,"tested code" comes down to volunteers who are willing to check that it works!

Cheers,

Erik


On Sep 22, 2013, at 10:59 PM, ms <devicerandom at gmail.com> wrote:

> On 22/09/13 22:30, Erik Lindahl wrote:
>> 
>> As we've discussed at a couple of occasions, many of the analysis
>> tools have way too many options that have only been tested in 1-2
>> special cases by the author, and when people test these on other
>> systems it frequently breaks. I would like to revert back to a case
>> where everything we present to users has been tested extensively, and
>> in a huge number of combinations.  Any code that has only been tested
>> in a special case does not belong in the public distribution, IMHO.
> 
> I am a gmx user (joined this ML because I needed to patch gmx long time ago for a project, still lurking), and please don't do that. Warning extensively that it is beta/poorly tested code is essential, sure. Removing it altogether is not so nice. It *might* work anyway, and even if it doesn't, people may still be able to patch it, by using it, noticing the break and remedy by themselves - and bringing the patch to devs.
> 
> Am I being naive?
> 
> thanks,
> Massimo
> 
> 
> -- 
> Massimo Sandal, Ph.D.
> http://devicerandom.org
> -- 
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-developers-request at gromacs.org.