3
$\begingroup$

There was a recent issue on Bioinfo SE here on a question about compiling gromacs for CPU or GPU (below).

It appears a question was deleted together with @MatteoFerla's response and one question remains. Having a valid response deleted would be frustrating.

The OP denies deleting the post, but it is possible there was cross-posting within the group by different users. For example, the problem is shared within the group, one OP posts here, a second OP posts here and one OP deletes their question.

Coincidence of two independent OPs seems unlikely on technical criteria.

I propose simple protocol where the responder has the right to comment that a discarded reply has occurred and request this is avoided. The OP can issue an explanation.

Is that reasonable? We have a similar unwritten protocol for cross-posting with Biostars.

I also propose @DevonRyan cleans up the comments below @MatteoFerla's comment(s) on the OP. That ok?

Technically the response is good and the question is fine. To answer the question one would need to look up the compiling flags, which is time consuming unless your the original developer. The technical issues between CPU/GPU wouldn't be apparent to a user: GPU parallelises very large numbers of very small differential equations, which a CPU can't do and thats how GPU speeds up the calculation. The reason you need to compile two versions is related to this issue.

$\endgroup$
3
  • $\begingroup$ To clarify, I had commented on the original post —asking which card and suggesting the same as the answer that was given and on the last post along with checking the CUDA toolkit was installed correctly (the heart of the problem). I did not hope to escalate this, just make sure that everyone was clear on what is good practice. So I am happy with the deletion of my comments. Thanks and sorry everyone $\endgroup$
    – Matteo Ferla
    Commented Nov 3, 2020 at 13:09
  • $\begingroup$ Your comments on the post are fine @MatteoFerla no deletion is needed. In context its good practice to point out the history of a question particularly when you've gone to some trouble looking up the answer (which got deleted). Lets see what Devon says. $\endgroup$
    – M__ Mod
    Commented Nov 3, 2020 at 15:32
  • $\begingroup$ @MatteoFerla For the record you comments were quite constructive and I think you nicely pointed out the exact cause of OP's problems. $\endgroup$
    – Devon Ryan
    Commented Nov 3, 2020 at 21:21

1 Answer 1

Reset to default
2
$\begingroup$

I see a related (now deleted) question by the exact same OP about trying to compile GROMACS with GPU support that is probably the one you saw. That one was lacking even more information than the more recent one, which is still lacking enough info to help OP but at least getting better.

I've cleaned up the comments on the post you noted since they're largely not relevant to the actual technical question (and were partially unhelpful. Please do note that down-voting on stackexchange is meant to be interpreted as "please work on this question more so we can help you", though I think in practice newer posters almost never actually interpret it that way (I would disable down-voting if I could, it does more harm than good).

Regarding your proposal for dealing with deleted posts we have two competing interests:

  1. We should endeavor to restrict content to answering the question.
  2. Context of replies is lost if one can't reference to other posts.

In general I think it's fine to reference deleted posts (possibly with a quick summary of their results). In an ideal world questions would just be rewritten entirely rather than deleted with new ones posted, but practically I think we all understand the utility of starting from a clean slate. So I would propose:

  1. Either OP or other posters to now-deleted content should feel allowed to reference said content and comment about that (e.g., "hey, sorry about the misunderstanding in the original post, hopefully this one is clearer"). We're all humans, we need this sort of thing to not start yelling at each other.
  2. These should be kept to a minimum, though, since it's not really answering the question at hand.
  3. These comments may get deleted by me (or another moderator) at some later point. I personally like to keep those around for a little while, since it helps with community building.
$\endgroup$
5
  • $\begingroup$ 1) Thanks much appreciated for the editing. 2) Sorry for my bad explaination. The idea was exactly how the comments were edited. If the OP deletes a post and thereby deletes all its replies and then makes an identifcal/very similar post (now without the previous reply), the responder has the right to point out what the OP has done and request not to do this in the comment section. The OP could offer a rationale explanation or apology in comments. $\endgroup$
    – M__ Mod
    Commented Nov 4, 2020 at 1:46
  • $\begingroup$ The gist is if the OP is taking the responder's goodwill for granted (which is what has happened), that is not cool and should be called out. However, just allow the OP some benefit of the doubt in case of a genuine misunderstanding. I'll clarify the post when I've a moment. $\endgroup$
    – M__ Mod
    Commented Nov 4, 2020 at 1:56
  • 1
    $\begingroup$ @Michael Thanks for the clarification, I've updated my reply (hopefully) accordingly. $\endgroup$
    – Devon Ryan
    Commented Nov 4, 2020 at 6:50
  • $\begingroup$ GROMACS seems extremely niche over here, such questions can be migrated to mattermodeling.stackexchange.com with no problems at all. Even questions asking for help compiling. $\endgroup$
    – user1271772
    Commented Dec 7, 2020 at 2:59
  • $\begingroup$ @user1271772 Thanks, I didn't know about that stack, I may migrate a few questions over there in the future. $\endgroup$
    – Devon Ryan
    Commented Dec 7, 2020 at 7:40

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .