Should there be a place to ask about programming libraries and technologies? [duplicate]

Question

Bjarne Stroustrup once mentioned in one of his bigthink interviews that one of the greatest problems most programmers have is not finding or using the right libraries.

On stackoverflow and programmers, I've observed a trend towards questions like "I want to do this, is there a [library | programming language] for that", or "I have these {n} [library | language] candidates for my project; which one would yield me the least amount of headache given these requirements".

Since questions like these aren't exactly programming questions in the sense of "this code should do this, but it does this, why?", they are often (though not always) being closed as "not constructive" because they do not have a definite answer (which is true, and according to the FAQ they would be off-topic on both stackoverflow and programmers).

Nonetheless, I think library questions are valid, and that it is helpful for any programmer to receive independent input from experienced programmers (at least it helps me a lot every time I browse through one of those closed questions, which I found through google). Does anyone agree that posting library choice questions should be explicitly allowed on either stackoverflow or programmers?

Answer 1

IMHO, there are two big problems with most of these questions, and they'll be troublesome on any SE site:

1. They're XY problems. Most folks asking for a library to send SMS messages aren't actually interested in a library - they just want to send SMS messages. And they probably have other specific requirements on top of that: must be able to send 900,000 messages per day, must be able to specify a specific number as the sender, must be able to continue working after numerous recipients call their phone companies with spam complaints, etc. I go into some detail about how to ask these in a manner that isn't a straight-up request for recommendations here: https://meta.stackexchange.com/questions/139399/what-exactly-is-a-recommendation-question/139404#139404

2. They're spam magnets. It's tempting to think of this as yet another case of one jerk ruining it for everyone else, but the truth is that Stack Overflow has reached a size where it is a tempting target for anyone with a library to promote. Sure, you can just search for "SMS" and post an answer to every question that happens to mention it - but unless you actually explain in detail how your library solves the problem being described, an honest-to-god problem description avoids giving you that veneer of respectability. And if you do go into that much detail, well...

Aside: although the examples are fairly SuperUser-specific, Jeff's classic blog post on Shopping Questions does provide some good, generally-applicable suggestions for asking them in a constructive manner.

Answer 2

I think these questions are genuinely very helpful indeed, and attract users to Stack Exchange.

It's important to keep things from being just a heap of opinions and make sure answers are based on facts and experience, so this falls very much into the category of things covered by good subjective, bad subjective. Stack Overflow can't cope with any kind of subjective, which is why it needs to be a separate stack exchange site.

This can work. For example, https://ux.stackexchange.com/ does a good job of voting up well-thought-through, detailed and justified advice above "that's really annoying" and "awesome, I want one" type answers.

I've made a proposal for programming-libraries.stackexchange at area 51.

Please suggest some good questions that will provoke good-subjective answers.

Answer 3

It's tricky to do, but you can in fact ask this type of question on Stack Overflow. The success of the question broadly depends upon where you are in the workflow of designing your project. If you ask too early, you're probably in one of the following positions:

• There are so many options, you want to narrow them down based on experiences your peers are willing to share

• There are apparently NO existing options. You want to confirm that nothing exists to avoid inventing another wheel needlessly

When it comes to library questions I personally feel that we should relax a little more. A modicum of research in the question itself provides breadcrumbs that could help others spend less time in the same sinkhole. What remains rather subjective is how much, in fact, constitutes a modicum of research in this context.

The first bullet, at least in the eyes of many users, falls into not satisfying the do your homework request that we ask of all question authors.

The second bullet basically attempts to document a very localized negative, or cycles back to the first bullet if options do in fact exist. No, nothing exists right now to do that. A year from now that might be downright misleading, so people edit and add options that have since emerged, and then we cycle back into the first bullet again.

If you ask at the right time, you've:

• Tried several libraries and narrowed down your choices
• Have a clear scope of want vs. don't want when it comes to how the ideal library works
• Have extremely narrow concerns that describe why you're still up in the air

If you're at that stage, you'll probably be in good shape. If you've read this answer and thought "this is just too damn complicated", then you've understood why I personally think that we should be a little more relaxed when it comes to libraries. Not much, mind you, but just enough to establish a clearer criteria on how and most importantly when to ask such questions.

Answer 4

FWIW I think we can make them fit here on SO, I think a lot of the problem comes down to whether people want them here.

The main problems with laundry-list questions are:

1. they seldom result in one definitive right answer
2. they require maintenance and curation
3. they can be rep-farming exercises where new users dream up an ideal list to ask for as a way of building easy rep
4. simply having a list doesn't mean it easily answers the questions of other users - they don't know what they want, remember?

We currently have a number of these types of questions and they are maintained (tolerated) for "historical reasons" (translation: there is still a reasonable number of users who want to keep them around because there are some really good answers in there).

I think the only way future instances of these types of questions can be productive is to:

• ensure they are CW so the question at least doesn't become a rep farming exercise (selecting an older version of the question and making it canonical is probably one of the better ways to start/keep a list like this)
• people need to take the time to maintain and/or merge new answers
• duplicates are ruthlessly closed; to a certain degree duplicates would be healthy as it helps address point #4 above (same question asked in different ways, helps those who don't know exactly what they want)

In any case, you will most likely still have a sizable portion of users who simply don't want these types of questions on Stack Overflow so you will need to win them over. I don't think starting a separate site is going to be productive as it will be quite low traffic and it won't prevent those questions being endlessly asked on SO.