How about explicitly stating that a MWE sometimes or oftentimes can identify the source of evil itself?
The original did contain something to that effect, but it was removed in edit 6 by @non-Jedi.
While there is a lot of useful advice that could be added to this post, I am still concerned that the longer we make it, the less of it will be read.
The post had been getting super long at that point, and I also removed a bunch of other stuff that might be considered helpful. The philosophy I employed was generally only to keep direct, actionable items and remove reasons/motivation and other commentary; the main reason that e.g. you should include a MWE is fairly obvious.
For what it’s worth, I have no personal objection to adding anything back in that people think is worthwhile, but I do believe that the longer the list gets, the less of it is actually read.
By the same token, the longer this thread gets, the less likely it is to be taken seriously by novices seeking help. How about splitting off the OP from the subsequent discussion?
Only the first post is meant to be read (the wiki part), the rest is discussion for that. Should we split it? I don’t know how wikis work on Discourse.
I agree with this. I think the items collected in this post (now a wiki, technically) convey the message rather well.
This is not meant to be a very detailed guide for every possible eventuality, just something that gets 90% of the common problems that prevent people from helping. FWIW, I am totally happy with people adding stuff, but even happier when they make it shorter.
My point was that this whole thread must look intimidating and confusing to a novice coming to the OP for advice. Threads have been split before, so I’m sure the top could be split off from the rest.
I am thinking about adding something along the following lines:
Instead of asking for translations of code in other languages (Matlab, Python, R, …) to Julia, it is recommended that you just explain what you are trying to do. This helps as it allows more people to respond (fewer people will know the other language well enough to understand what you are asking for); also, not all constructs have a direct mapping to Julia so context is useful.
I think this is a good point, but I might say “instead of just asking… it is recommended that you also explain…” Or something like that. I 100% agree with the sentiment, but having example code in a different language can often be helpful.
Then again this might undermine the clarity of the message, so feel free to disregard.
This seems like a specific kind of XY problem. Maybe better to just include a brief point requesting that they provide context for their question–what and why they’re trying to accomplish in an overall rather than local sense?
I’ve also been thinking: would it be better to simply link to something like Stack Overflow’s “How do I ask a good question” and limit this guide to things that are specific to Julia/this forum?
In principle, linking to a guide like that could work; in practice, I am not sure that the people who need it most would click through 2 links.
I agree. And also, though I think that post has some useful guidance (and could be linked to at the end as “further reading” or something) your post has a bunch of additional julia- and discourse-specific stuff that I think is really valuable.
we should have a database of FAQs resolved and link new users if they have similar questions. in many cases, similar questions are asked and we don’t have a compilation of answers in one link to make the interaction faster and avoid redundant replies/troubleshootings.
Hmm I think that the search function and the “relevant topics” feature when creating a new post covers that already. I am always a bit reluctant to maintain multiple source of information.
Not sure if someone who does not utilise these existing features will find their way to an FAQ
curated FAQs are always helpful as it indicates the most commonly ask/popular/critical questions. users can be linked to the FAQs as an entry point and if it’s properly curated, it can lessen the traffic of questions and also easy to cite it. of course, maintaining it can require lots of work but it can be done with some bots/automations too.
Yeah I know the benefits but I doubt that this will change anything compared to the maintenance work needed to keep it in shape. We have similar “information source” issues in our collaboration for many years and it seems that some things will never change. Whatever we tried, newcomers behave the same no matter what fancy tools you provide. However, I don’t want to be too pessimistic It feels like there should simply be a solid base of patient and polite experts who react correspondingly.
we like automations and lazy to be redundant that’s why we like to program tasks. if questions are repeated many times, it makes sense to collate them like in a datascience way, else, we are stucked in endless manual loop
i miss the chatbot in debian which stores lots of helpful tips and solutions which can be triggered by simple keyword. anybody asking same questions and a factoid exists, you can just trigger the bot to display the factoid to the OP.
something like this: https://ircbots.debian.net/factoids/search.php?q=install&term=key
I am not sure if you are aware of this, but the Discourse engine has a similar feature: if you start typing up a new post here, it will suggest previous ones.
In any case, I agree with @tamasgal about diminishing returns of putting a lot of q&a in FAQs. The manual has a FAQ, but it seems to serve the same purpose as FAQs everywhere: to save typing up the same answer again and again. Contributions to it are usually welcome, as PRs.