An assortment of tips for mentoring
One of the biggest helps to mentoring can be to have a file to hold notes for each exercise you mentor. You may find that many solutions can benefit from the same suggestions, so, by keeping notes, you don't need to keep writing up the same suggestions from memory. And, by having the suggestions in one place, you can keep refining them over time to make them clearer.
If you're not sure how to get started with your notes, you may find a mentoring.md file for your track's
exercise under exercism/website-copy/tracks.
If it exists, it may include examples of reasonable solutions, along with common suggestions and talking points
to prompt further discussion.
If it doesn't exist, you may want to go back and create one after you've made your own file of notes for that exercise.
Also, even if you only mentor one language now, you may mentor more in the future. It may help to organize your mentoring notes by track as well as by exercise name, as different tracks will likely require different suggestions for the same exercise.
Mentoring notes are handy, whether you mentor the exercise frequently or infrequently. If you mentor the exerise frequently, it saves a lot of typing from scratch, when you can just copy-and-paste from your notes. If you mentor the exerise infrequently, it can remind you of suggestions to make that you may have forgotten in the weeks or months since you last mentored it.
It's okay for mentoring notes to differ between mentors. Here is one way to structure them, but it is not the only way.
Congratulate the mentee on passing the tests (if they passed them).
If the exercises has been sitting in the queue for a few days, maybe address that with something like:
Sorry it took a while for someone to get back to you. There is currently a shortage of active JavaScript mentors for
Resistor Color Duo.
Itemize what you like about the mentee's solution. For example:
I like this solution is succinct and readable.
I like the use of indexOf.
I like this uses the (first * 10) + second approach to avoid casting between number to string back to number.
I like this does not use looping/iteration.
I like the destructured parameter.
Next could come your frequent suggestions.
It can be very helpful for the mentee if a link is provided for each new language feature you introduce. For example:
It's not necessary for this exercise, but perhaps consider converting the function to an arrow function.
Although we don't want to give away the solution, sometimes a mentee learns best by example. To put a snippet of code in a collapsed details section can provide that example, which the mentee can choose to expand or not. For example:
<details><summary>Spoiler Example</summary>
<pre>
export const decodedValue = ([firstColor, secondColor]) => COLORS.indexOf(firstColor) * 10 + COLORS.indexOf(secondColor)
</pre>
</details>
Toward the end of the notes you might include a link to a published solution which represents the suggestions in full.
At the very bottom of your notes you may want to put extended explanations that mentees sometimes ask for. These explanations don't come up often, but it can still be good to record them the first time you use them, so the next time, which could be weeks or months away, you won't have to come up with the explanation from scratch. For example, sometimes a mentee will ask about how the multiplication approach would work for Resistor Color Duo if black was the first band for a leading zero:
Black for a first band is a good point to consider, so let's consider it. The resister color is meant to represent the amount of ohms for the resistor, and a leading zero would not be used for a multi-band resistor. So black would not be a first band. Besides,
parseIntorNumberwill also remove the leading zero.
An optional cateory of data to keep in mentoring notes is a record of benchmarks for various solutions or approaches.
A common concern for mentees is how performant their solution is. This is especially the case for "lower-level" languages such as C, C++, Go, and Rust. Along with how idiomatic their code is, mentees of other languages are also often concerned with the efficiency of their code.
Benchmarking is not something that a mentor is expected to do. However, mentees are often particularly impressed by how a benchmark of their solution compares with other approaches.
Go is a particularly friendly track for benchmarking, as benchmarks are often included in the test file. Other languages may require some research to determine what method would work best for you. For instance, if you only use the online editor, then you would be looking for a place to run benchmarks online. For example, JSBench.me is an online benchmarker for JavaScript.
If you run code locally, then you have the option of downloading benchmarking software you can run on your machine. For example, Rust can use Criterion, or cargo bench with benchmark tests.
There are at least a couple ways to keep track of benchmarks. One way is to keep a running list of all the ones you benchmark, but that can get unwieldy if the list gets long. Another way is to keep a list of repesentative benchmarks for different approaches. Mentees often want to see the code for the faster approaches, so if a faster approach is published, it will likely be much appreciated to provide the link to it.
If providing a link to a solution you benchmarked, be sure to provide a link to the published solution and not to the mentoring session. Not all solutions that are mentored get published.
There may be some features of the language that you find addressing for more than one exercise. When about to copy-and-paste a suggestion from one file to another, perhaps consider putting it into its own file instead. Again, a benefit for keeping a suggestion in one place is to make it easier to refine over time. It also makes it easier to find when using it for an exercise you haven't used it before. Rather than trying to remember in which exercise you addressed the suggestion before, you can go right to the suggestion's own file.
Mentees are encouraged to specify what they expect to get from the mentoring session. They will often express that in the form of a question. If the question is something for which you don't know the answer and are not interested in, it is okay to leave the mentoring request for another mentor.
If you don't know the answer but want to find it, then it may be best to not pick up the mentoring request until you've learned the answer. If the mentoring request is gone by that time, at least you've learned something and didn't make the mentee wait.
One exception to this may be if the mentoring request has already been in the queue for several days or longer. In that situation you may want to pick up the mentoring request and give what feedback you can, and let the mentee know you will get back to them on their question. Of course, it is important to follow up on that, either to inform the mentee of the answer, or to let them know you couldn't find it. If you couldn't find the answer, it may be helpful to the mentee to describe what ways you took to try to find the answer. The mentee may respond with other ways to try to find the answer. Between the two of you, the answer may be found.
If you have exhausted every way you know of to find the answer, you can suggest the mentee end the discussion and resubmit their request on the chance that another mentor may provide the answer. If they care to, the mentee can post on the ended discussion to share the answer with you once they learn it. And likewise, if you learn the answer later, you can go back to the ended discussion and let the mentee know.
If you do know the answer and want to address it, a good place to do so is in between telling the student what you like about their solution and offering suggestions for other approaches.
Code can fail either because it does not pass all the tests or because it doesn't compile or satisfy the interpreter.
Different mentors will have varying inclinations and/or patience for dealing with failing code, which may somewhat depend on how it is presented, as failing code is not always presented in the same way.
Sometimes a mentee will try say that they tried another approach and it didn't work, and they will ask why it didn't work. The code may not even be provided, or it may be posted in a practically unreadable comment instead of in an iteration.
A solution tested in the web editor can only be submitted for a mentoring request if it has passed all the tests. One of the reasons is so the mentor can focus on suggesting improvements or other approaches to the existing working code. Debugging code is not necessarily something a mentor wants or is expected to do. However, a failing solution submitted through the CLI can be submitted for a mentoring request, with the student asking for help to solve it.
If the failing code has not been provided, and the described failing approach does not sound like a good one, it may be enough to suggest that, instead of using the failing approach, another approach could be one that is neither the failed approach or the one they used that did pass. Or it may be enough to explain why the approach they used is better than the failed approach, without getting into the details of what bug was in the failed approach.
For example, a common occurrence is mentees having trouble with Robot Name. Either the tests time out or they fail to generate enough names, and they want to know how to fix it. If you have the inclination and the patience, you can certainly analyze their code and suggest how to address the problem. Or you can explain that checking randomly generated names causes more collision as more names are generated, and suggest that another approach could be to generate the names sequentially and then shuffle them.
If the failing code has been pasted into a practically unreadable comment, you may want to give what feedback you can on the passing solution, and suggest they submit the comment code as another iteration. You can also suggest that the mentee then check the errors for the failing iteration as a guide to where the problem is.
If the code is in a failing iteration, then it can be helpful to direct the mentee to check the errors for the test run. Some languages need a bit more guidance on how to read errors or test results than others. It may be helpful to quote one or more parts of the errors and explain to the student what is meant.
Ultimately, it is not the mentor's responsibility to fix the mentee's failing code, but the mentor, if they want to, can suggest ways to the mentee for fixing it themselves.
It may be that you sign on to mentor a track, but you never see any exercises in its queue to mentor. You may think that something is wrong, but there are at least a couple reasons for this. One reason is that people may not be requesting mentoring on the track for now. Sometimes a track may have periods of inactivity. Another reason is that other mentors may be picking up the requests before you see them. This is likely to happen for a popular track that has many active mentors.
If there are many requests in the queue, there are a few ways to approach mentoring them. You may want to work from the oldest to the newest, so those who've waited the longest are dealt with first. Or, you may choose to work from the newest to oldest, especially if the oldest have already been waiting a long time. That way, people who have been recently active don't have to wait for the backlog to be handled.
If there are multiple requests for the same exercise, you may want to work through them in batches of the same exercise to maintain focus, as opposed to going from exercise A to exercise B back to exercise A.
It may be that a request for an exercise that you are not interested in has been sitting there for days or weeks. You can choose to not address it in hope that another mentor will pick it up, or it may serve as motivation to give the exercise a try yourself. One thing that may be helpful is to look at the submitted solution. It may use an approach that you hadn't thought of, and that approach may make solving the exercise more attractive to you. But, if you look at the code and are still not interested in solving the exercise, there is no harm done. Just because you look at a mentoring request doesn't mean you have to click the "Start mentoring" button.