Version tags considered harmful

Question

We've noticed that a huge percentage of questions are tagged drupal-6 or drupal-7.

This is a dangerous pattern for your community to fall into. Here's why:

1. Version tags are redundant. The site is already called "drupal". We don't need to know that it's drupal-6, drupal-7, drupal-8, drupal-9 in every single tag on every single question. It's redundant.

2. Version tags are a crutch. Because these tags dominate the site, and contain the name, they are the first thing users will turn to when tagging their questions. This means users will fail to tag questions with any other meaningful tags, since "I already tagged my question with drupal-6!" This isn't hypothetical. We've seen this happen time and time again. Version tags are a mindless replacement for thinking about what your question is about. "Oh, this question is about drupal-7. Done."

3. Version tags make questions disposable. The entire point of Stack Exchange is for questions to be editable, timeless resources -- a version tag implies exactly the opposite and ties questions to specific moments in time, with no incentive to improve them to be relevant to future versions. This is extremely dangerous!

4. Version tags encourage needless question duplication. Rather than "How do I do X?" which can cover both supported versions, now there has to be "How do I do X in Version 1?" and "How do I do X in Version 2?"

5. Version tags aren't necessary for most new questions. New questions are very likely to be about the current version of the software -- so for all new questions, tagging with the current version is usually not helpful.

On top of that, how many versions (outside of freakish "I still use Windows 2000" edge conditions) will really matter at any given time? Two at best? And certainly any OSS project worth its salt will strongly urge people to regularly migrate to the latest version as soon as they can, so that the community can move forward to bigger and better things.

Thus, here is what I propose.

• Eliminating the version tags on existing questions. We can do this in one click globally.

• Treating version tags as the exception not the rule on new questions -- if you have a strong case to make that the question you are asking is ONLY relevant to one version and can NEVER EVER be relevant to any other version, then -- and only then -- add the version tag.

• If you encounter an old question that is truly specific to a version, and cannot be edited to be version-agnostic, retag it with a version tag.

The bottom line is that version tags have to be treated as the exception, not the rule. It is important to fix this now while your community is young? Otherwise you are creating some deep problems in your community that will be very hard to fix 1 or 2 or 3 years from now.

Answer 1

This might all be true about many other versioned things, but in case of Drupal, the difference between 6 and 7 is huge and the answer depends in many cases on the version used.

Drupal 7 was in many areas a major redesign, many things just don't work like they did in 6. The same things will likely happen with Drupal 8. Drupal core isn't backwards compatible

1. We already had a similar discussion if version-6 should be used instead of drupal-6. We went back to drupal-6/7 however, because these are the official terms (Nobody talks about Version 6, but always "Drupal 6") and some contributed modules also have different major versions where a differentiation might be useful.

2. It is IMHO important that users choose one of these tags. Because a large amount of the questions that do not have this tag initially have a comment that asks which version they are using so that it can be answered properly.

3. While this is true, this is not how Drupal works. Drupal versions and "ways of doing things" are not timeless, they often change together with new core versions. One example is the new entity/field concept which allows to add various types of "fields" to entities (nodes, users, comments, ...) while in Drupal 6, this was only possible for nodes and was not in Core but a contributed module. Many modules exist to work around that limitation in Drupal 6 which are simply not necessary anymore.

4. As explained above, chances are pretty high that the answer to these two question is different based on the version used. Maybe at some point Drupal core will stabilize and such a differentation wouldn't be necessary anymore but it is also very possible that at the same point new frameworks/systems will appear and Drupal's rise will stop and it will eventually vanish.

5. Just look at the numbers of how often those two tags (drupal-6, drupal-7) are used. We have 627 Drupal 6 questions vs. 543 Drupal 7 questions. (Drupal 7 was released in Januar 2011). So it's pretty equal and it would be wrong to assume that new questions are for Drupal 7. That again might be Drupal specific, but because of the vast amount of changes and the fact that contributed modules often lag behind a bit (and you usually use a lot of contributed modules on every site you build with Drupal), the shift to the next major version usually takes a bit of time (People still work on improving existing D6 sites, some new projects still might use Drupal 6 right now because an important module they're depending on might not yet be available for Drupal 7 and so on).

That said, I totally see what you are saying and that it might/will be a problem in 1/2/3 years. But not using version tags right now is just not possible/doable.

Maybe we can find a different way to specify the used version and/or if a question/answer is version specific or not.

Answer 2

Trying to give some examples for questions that would be answered more or less differently based on the core version. This is too long for a comment, so I'm doing it as a new answer.

Maybe this can help to explain what I and kiamlaluno mean...

Q1: How can I relate two nodes together?

D6 A: With the CCK project, you can add fields to nodes. Make sure to enable the Node reference module and then add a Node reference field to the node type you want to connect to another.

D7 A: The ability to add fields to nodes (and other things) has been added to Drupal 7 core. However, the node reference module did not make it into core and is now part of the References project. There is also a new project called Relations which aims to provide a generic solution to relate entity X (node, comment, user, ...) to entity Y. Additionally, that module makes the relation between two entities also an entity so you can add information/fields directly to the relationship between two things.

Explanation: The answer is similar, but you need different modules in D6/D7 and even within a core version, new contributed modules/ideas are developed which can change how things are usually done.

Q2: How do I load a user with a specific name?

D6 A:

<?php
$account = user_load(array('name' => 'user name'));
?>

D7 A:

<?php
$account = user_load_by_name('user name');
?>

Explanation: user_load_by_name() is a new helper function, see http://api.drupal.org/api/drupal/modules--user--user.module/function/user_load_by_name/7, that does not exist in D6 and uses the new function user_load_multiple() that did not exist in D6. Using either code in the other version would not work.

Q3: How do I load a user based on the user id?

A

<?php
$account = user_load($id);
?>

Explanation: This time, the answer is the same and does not depend on the used version (But it could theoretically change in Drupal 8 or 9, who knows). Also, in Drupal 6, it was also possible to pass an array to user_load() like this: user_load(array('uid' => $id)), this does not work anymore in Drupal 7.

So, how would one know that the answer for Q1 and 2 is a bit different, but not for Q3?

The question now is, how to de expose that information without misusing the tagging system? I honestly have no idea...

Answer 3

Jeff, you and kiamlaluno are making such a mess of this site. Please stop making rash changes like banning tags…

You may have different ideas, but everywhere in the documentation on Drupal.org, it's tagged with what version it pertains to. If that causes problems with the StackExchange software, perhaps it's the software we need to fix?

If we were to build a QnA site in Drupal, we would just have the version as a separate attribute instead of tags, but since that's not possible with SE, tags will have to do. Versioning info is so important that the QnA is useless without it.

Examples are fairly easily translated from one version to another, but you need to know what version the example code is, to realise that you should expect an object instead of an array or whatnot.

Answer 4

1. I agree that using tags containing the word drupal is not helpful, as the site is about Drupal. Others reported that "Drupal 7" is the way a Drupal version is reported, but that has nothing to do with how a tag should be named. If the problem were users keeping to use drupal-7 instead of version-7, the first could be made a synonym of the latter; in that way, no tag used in questions would contain "drupal."

2. It seems it is already happening, as some of the recent questions just contain the tag for the Drupal version:

   • How do I change a MySQL query into a Drupal query?
   • https://drupal.stackexchange.com/questions/3968/the-specified-date-is-invalid-error
   • https://drupal.stackexchange.com/questions/3967/why-admin-can-not-create-new-content
   • My implementation of template_preprocess_block() isn't working

3. I don't think that questions for Drupal 5, for example, are restricted in time; it is still possible to download Drupal 5 and, for what it matters, Drupal 4.7. The fact they are not supported anymore means only that bugs will not be fixed anymore, in the same way there are things in Drupal 7 that are not changed anymore; it does't mean Drupal 5 could not be used for a website.

4. It is possible that the same question is asked for Drupal 6 and Drupal 7, in the same way similar questions are asked about Perl 5 and Perl 6. Supposing there is already a question about Perl 5, what should a user who is interested to know the answer for the same question for Perl 6 do? There are many things that are changed in Perl 6; for example, in Perl 6 is not possible to use __PACKAGE__ anymore; where Perl 5 uses foreach my $x (@whatever) { … }, Perl 6 uses for @whatever -> $x { … }; where Perl 5 uses $str =~ m/^\d{2,5}\s/i, Perl 6 uses $str ~~ m:P5:i/^\d{2,5}\s/.
   The difference between a programming language, and a CMS is the time between new releases. How long did we had to wait before to pass from PHP 4 to PHP 5? How much time will pass between the first release of Perl 5 and the first release of Perl 6?

5. Considering that the current Drupal version (as defined in drupal.org) is Drupal 7, taking the assumption that new questions are about Drupal 7 is a wrong assumption. There are 627 questions tagged drupal-6, 573 questions tagged drupal-7, and 5 questions tagged drupal-5. All those questions have been asked when Drupal 7 already had an official release (which means any release that is not a development snapshot).

To make a summary, I am not against removing the Drupal version tags, but that will not stop questions for which the user is interested in how to do something using a specific Drupal version, in the same way nothing has stopped users to ask a question about the regular expression to use to resolve a specific task, instead of making the question as much generic as possible.
If we are going to remove those tags, then they should be blocked; questions about Drupal on Stack Overflow are using those version tags, and we would need to avoid the tags are imported into Drupal Answers.

Answer 5

Version tags are redundant. The site is already called "drupal". We don't need to know that it's drupal-6, drupal-7, drupal-8, drupal-9 in every single tag on every single question. It's redundant.

This is not true. To answer a question properly we need to know what version of Drupal is involved. Drupal 4, 5, 6 and 7 are vastly different. Any answer beyond the most basic given to Drupal 4 or 5 would be completely useless for a Drupal 7 site. What you are saying is like saying We don't need to know what programming language is used, all we need to know is that it's about programming for the SO site. I know it's not exactly the same thing, but what I'm saying is that Drupal versions are vastly different and an answer for one version could be outright wrong for another one.

Version tags are a crutch. Because these tags dominate the site, and contain the name, they are the first thing users will turn to when tagging their questions. This means users will fail to tag questions with any other meaningful tags, since "I already tagged my question with drupal-6!" This isn't hypothetical. We've seen this happen time and time again. Version tags are a mindless replacement for thinking about what your question is about. "Oh, this question is about drupal-7. Done."

I can see how this can be a problem

Version tags make questions disposable. The entire point of Stack Exchange is for questions to be editable, timeless resources -- a version tag implies exactly the opposite and ties questions to specific moments in time, with no incentive to improve them to be relevant to future versions. This is extremely dangerous!

You cannot make timeless answers for Drupal for many reasons. The API changes a lot, the modules used changes as well, and standard ways of solve certain problems is created. My guess is that about 90% or more of the answers created thus far is completely irrelevant in 4 years, since Drupal will have evolved. Not tagging questions can fix this, you would have to stop the development of Drupal core.

Version tags encourage needless question duplication. Rather than "How do I do X?" which can cover both supported versions, now there has to be "How do I do X in Version 1?" and "How do I do X in Version 2?"

This is both true and false. It's true that you would have to create the same question for Drupal 6 and 7, the alternative is to create an answer for each version with some answers. IMO this is much more problematic. You can't tag answers so you have no other way than writing it in the text field that an answer is related to this and this version. Over time such a question would have many different types of answers, some to one version or the other, some to multiple versions. The number of upvotes can be related to which version was used most at the time etc. So when a user is redirected from Google, how should he be able to figure out which answer is useful. He need a Drupal-9 solution, but since there is no way to filter the question or answer he would have to read through all the answers making the voting system useless.

Version tags aren't necessary for most new questions. New questions are very likely to be about the current version of the software -- so for all new questions, tagging with the current version is usually not helpful.

This is true maybe 1/2 of the time. Drupal have a very long period where two vesions are used equally, since a lot modules which is used all the time, need to be ported to the new version, which last time took about a year or so.

I understand some of the dangers and problems that can arise by using version tags, but it seems to me that not using them will create a lot of problems as well, only different ones. For this site to succeed, we need a way to be able to filter on version. It would make more sense to "tag" an answer with a version tag and not the question. This would be a great feature to allow tagging answers with a small selection of tags, but this is not possible atm. So we only have one thing to do, and that is tagging the question. The alternative is to write the Drupal version in the title of the body, or figure it out by asking in comments, which would slow down the whole answering process.