7
$\begingroup$

Every now and again software maintainers and developers add new features and fix bugs. Sometimes these are smaller changes, which do not break anything (minor revisions), sometimes these introduce changes in input syntax, api, etc., which might break existing workflows (major revision).

For example, when switching form xtb 5.9 to xtb 6.0, the interface was completely changed. Command line options were now accessible with double dash instead of single dash. Obviously a user reading answers here is expected to abstract from the (maybe older) information to the current one.

Tags for the major software packages have been established, and I think they will serve well for the time being. Planning a bit ahead though: Should we create tags for specific versions of these packages?

There are plenty of reasons, why this might be a good idea. A counter-argument is that this might create a huge pile of tags you can hardly keep track of. For example:

  • Unfortunately there are still plenty of researchers out there not having access to the most current distribution of their software. There are still researchers using Gaussian 09 predominantly, even though it doesn't have support any more. Tagging will immediately give the community a feedback, which version of software is used - an information, which is surprisingly often not as easy to come by as one would think.
  • Over time this Q&A will age and some software specific information will no longer be valid. Screenshots may change, interfaces renamed, options restructures, etc.. While I believe most of you answering these questions will include this information in the body, it might be worthwhile having something to sort by.
  • Occasionally someone needs to go back to an older version to replicate results. For those it would help tremendously identifying questions with a specific old version in mind.

I'm sure you can think of more.

So I propose and instead of , or and instead of , or and instead of , etc.. Obviously for some packages, this has to be more nuanced. What do you think?

| |
$\endgroup$
5
$\begingroup$

Someone else might have a more relevant opinion than me, since I do not actively use the three software suites for which you propose introduction of new tags: ORCA, VASP, and Gaussian.

I know that Psi4 is removed enough from Psi3, and ACESIII is removed enough from ACESII, almost as if these can be considered different software. We don't have any Psi or ACES questions yet though, so it wouldn't yet make sense to make separate tags for them.

For Gaussian, VASP, and ORCA, perhaps we can wait until we get answers from people who are really active in those tags.

| |
$\endgroup$
  • $\begingroup$ In the same sense as Psi3/4 I would say Gaussian would fit in a similar category. There are plenty of features which are in 16, but not in 09 and earlier. For Orca I don't know the specifics, there have been some breaking changes from 3-4 and I expect the same going further. I don't think it would hurt though... But you are right, let's get some more opinions. $\endgroup$ – Martin - マーチン May 14 at 9:33
2
$\begingroup$

I agree that it would be good to introduce these tags. As the SE matures, we will have higher-level questions and since this is the SE that deals with development of matter modeling codes, these version specific tags will come in handy. I have had to move between versions of some codes to get certain things done.

However, I think that "instead of" these tags should be "complementary to" the generic code tags. For example, questions specific to Quantum ESPRESSO should be tagged as and then a second tag for example. This should be so that we don't lose the generality of the questions, just as they should use the and other general umbrella terms like before or something that specific. The key is to attract both the -broad- community who can provide insight, and the -experts- on a specific topic, including the developers themselves! This is actually already happening on the site! :)

Short answer to Should we create tags for specific versions of these packages?

YES

| |
$\endgroup$

You must log in to answer this question.

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