I did a "365 days of code challenge" in 2019. I logged my progress everyday: Day 1-100, Day 101-200, Day 201-365. After I completed the challenge, I continued to log in my 2020 coding log. This is my 2021 coding log.
- Using Github projects
- It's taking me a year to launch. Am I doing everything wrong?
- Security vs. Shipping fast
It's been over a month since I last posted, but I have been working hard.
I started tracking my progress on a github project: Wishtender to launch.
Github projects helps me prioritize which tasks to attack next and helps me track my progress.
Since I started using Github projects on May 14, here's the to-do's I tackled:
- limit failed login attempts
- rate limit app
- make sure sessions delete after logout
- session secret needs to be moved to env file
- add Delete account view
- add change-email views
- migrate changes necessary for react-hook-form 7
- add Change-password views
- add Reset password route
- add change email route
- add soft delete user end point
- add hardDelete in userService
- Add progress bar to change handle
- Prompt to add wishes
- Edit information missing way to edit aliasName
- Price should populate with ".00" or work when no ".00"
- Price doesn't show after login
- update wish photo not working
- wish stays in cart after deleted or updated from wishlist
- image gone after delete wish
- Forward to SetUp after logging in with an account that hasn't created a wishlist
- Should render "Your cart is empty"
- reply button visible when already replied
- Currency dialog needs to be styled
- edit gift had wrong background image
- remove currency = "USD"
- style styleddialog mobile
- Currency selection in app bar
- rates api cache
- ratesapi.io shut down
- Currency dialog
- style "add-to-cart"
- Style activate button
- style connect success page
- Style "successful order" page
Script to get to-do's
// run in browser console
const itemsArray = document.querySelector("#column-cards-14322832").children
[...itemsArray].map(i=>i.children[1].children[0].children[0].children[2].innerText).forEach(p=>console.log(p))
This question has been on my mind. I posted on indie hackers:
It's taking me a year to launch. Am I doing everything wrong?
The answers clarified a lot. I now see I over estimated the level of quality people mean when they say "build an mvp". Hearing real examples of buggy, hacky, insecurely launched apps helped.
I also started reading The Lean Startup to better understand mvps and The Mom Test to understand my customers.
I've switched my focus to security and away from some of the other issues in my app. Once I establish minimum security to protect my users real identities, I will launch with one user and slowly add more. "Big reveal" launches are not helpful for a solo founder, and they really warped my perception of what launching should look like.
Still, I was confused about the suggestion by several indie hackers to skip security. So I posted:
I learned the blatant suggestion to skip security is short sighted. Posts on the indie hackers about getting hacked are evidence of this.
In my case, security is a necessity. You need to risk assess for your use case. For instance, my app needs to protect the identities of the users. Therefore, preventing injections is important. On the other hand, I don’t need to prevent DOS attacks until they appear. One DOS attack isn’t going to harm me or my customers. However, if you had an app that needs to run 24/7 because it monitors emergency situations, you would need to do everything to prevent DOS. One DOS attack could cost you a customer, or worse, a life.
You could also build your app so that if it got hacked, it wouldn’t matter. For example, when my user gets a notification on my app, instead of my app emailing the user, it could email me. Then I could open my paper notebook, find their email, and manually email the user their notification. Thanks to my manual solution, I don’t need to store sensitive information on a database. If my database gets hacked it wouldn’t matter. My users' emails are not on there.
This week I will figure out what other security measures are a priority and learn and implement those.
WishTender Reflections
Todays Post:
I've been working hard on WishTender. You can see my progress in the issues tabs of the frontend and backend repos.
In this journal entry I explore the balance between "learning in preparation" vs "learning when necessary".
5/4/21 am
I might have spent too much time learning when building WishTender, instead of doing. But it’s really hard to say. For example, did I spend too much time learning about error handling? Today it feels like yes. Because I can see now that it’s very easy to add error handling or change error handling later. And you don’t really understand how errors should be handled until later on in the app process. Meaning: you learn their importance by experience . However, is this accurate? Maybe I’m blinded by my current knowledge. It’s easy to think “oh it’s easy to add error handling” because I know about it. But maybe had I never learned by reading about error handling I would have never fully grasped it through experience alone. I wouldn’t be here thinking “this is easy why did I spend so much time on it?” I would be thinking “I wish I knew more about error handling” you can’t know the importance of a concept before you learn it. And it’s even difficult to look back and know the importance in hindsight.
I’m often worried that there are somethings I need to learn now because they will be complicated to correct later. How often is that the case? It seems rarely. But the next question is, if you build something incorrectly how much will you have to learn to correct it? And will that happen at a inconvenient time, for example after the product is launched.
Ex: I worry about database changes. I believe referred to as migrations. I have never done a database migration. Well I did in rails. But that’s like built in. Also that was very long ago. So if I need to change the database, well now I’m dealing with real customer data that has to be transformed. And I will need to learn how and then do it without destroying data
In this journal entry I explore how I over estimate other peoples expertise and underestimate mine. And how that is a disservice to both myself and the people i'm putting up on a pedestal.
5/10/21
Building wishtender I learned:
I trusted people who I viewed as more experienced than me too much. Example, one of my friends has a successful business. He asked me months ago why I don’t just hire a person from India to build wishtender. At the time I had this intuition that it both sounded a bad idea and impossible. How will you hire someone in India to build an app for $500? Will Indian devs actually do that? And if they do will it be good? But I didn’t ask my friend these questions. I shoved these intuitions away and I just assumed he knew what he was talking about- that it was possible and an ok idea to hire an Indian dev to build an application for $500. I invented other reasons why I wouldn’t do it - “I want to learn myself”. I trusted my friend even to the point that I believed maybe I was making a mistake by not outsourcing for cheap. But the real reason I didn't outsource was because of my intuition deep down that this was a bad or impossible idea. It turns out my friend didn’t know what he was talking about. Experience in hiring a dev to build your Wordpress e-commerce site does not give you experience in hiring devs to build an application. This became apparent recently. Fast forward to a few months later I talk to my friend again about how long it’s taking me to build wishtender- longer than I expected. My friend says I should hire his dev because his dev is the best he’s used and can do anything. So I finally contact his dev. But his dev has terrible communication skills and I never even get to find out if he can program in react or node because he never answers my direct questions. I talked to my friend and asked what was up with his dev? My friend said his dev was not great at communication. As we talked it became clear to me that his dev was great for his purposes but not for mine. It also became clear that my friend didn’t know there was a difference between building a Wordpress site and developing an application. And I now doubt his dev would be able to work on wishtender because he did not have the right skills.
So it turns out my intuitions were correct. But I was too embarrassed to acknowledge them. I should have confronted my skepticism immediately and asked my friend more questions so I could learn that his advice was not appropriate for me. Even though I didn’t act on his advice, and instead continued to build wishtender myself, by not confronting my skepticism I left doubt in my mind.
I don’t fault my friend for his overly confident advice, he didn’t know that it wasn’t applicable to me. I do fault myself for not speaking up. I could have helped myself by voicing my skepticism. It would have helped my friend too by teaching him the difference between applications and Wordpress e-commerce sites. This knowledge could have helped him one day too, maybe if he decides to pursue a software project. It was my responsibility to teach him what I knew about software. But by downplaying my expertise I was unhelpful to both of us.
When someone talks to you they don’t just want you too believe everything they say, they want value from you. And value can be in the form of sharing your doubt.
WishTender Notes
Debug safari mobile on mac
If still issues try:
You have to turn on the bluetooth and personal hotspot. Connect both using the bluetooh 'Connect to Network' and you can see your iPhone on the 'Develop' tab...
For me personal hot spot over USB cable worked.
Sometimes I have to disable WiFi on my phone as well.
Then connect to localhost from safari mobile
If there's CORS involved in the app, make sure to add your domain to the origin for the cookies. Which is probably the same as the IP found in the article above, also found with req.get("referer")
WishTender Update
Todays Post:
Journal entry where I talk about
- Estimating duration to complete the app
- What's needed for an MVP
- Where do I get my passion for WishTender?
How do you know how long things take?
I don’t know if I will launch wishtender on the 20th. How do people figure out how long this takes? Deploying took long. How would I know how long it was going to take? I didn’t know about how complex cors gets when you deploy. There’s so many things I underestimated the complexity of. So how do mvps normally work? Even the bare minimum mvp still needs backend validations etc. And without tests things get hard to fix so it seems like you need tests too. And so on. What do you not need in an mvp?? Maybe it depends on the project.
I think my project could turn into a real profitable business. I didn’t follow my passion. I just looked to bring people value in a topic that is sort of fun for me. I love clothes and gifts and fun products. I think I like sex workers because they don’t give a damn about their reputation which is cool to me. But I don’t know many too well. And I’m not super sexual myself. But I like cool women who are hot and confident and dress cool or pretty so I may find some of them working on wishtender.
I’m not passionate about sex workers I just think I’ll find them and their gifts fun so the business will be full of color for me. My passion is lucid dreaming and I hope projects like wishtender will help me get rich so I can fund lucid dream projects.
I also am passionate about entrepreneurship. So I like the challenge of building something useful for people. Elon musk funded space x. But he started with eBay(wait was it eBay or PayPal?)- fun but not as cool as outer space. But it gave him the money to build space x. That’s how I see what I’m doing. Wishtender is my eBay (wishtender or whatever project I succeed at). Lucid dreaming is my passion.
I think some people maybe are surprised I was starting a business catered to sex workers. But I don’t care. I don’t have any natural inclination to judge them like other people do. I just have goals. I will do what I need to do to achieve them. Being non judgmental is my strength. When judgements get in the way of your own goals- that’s no good! Because I respect sex workers, I can build a business for them, help them, and fund lucid dreaming advances. Well, if all goes well. My respect for them has nothing to do with liking porn. I think porn is bad for health with some caveats. But I think it’s not my business to judge what commerce people partake in. I’m supportive of people enjoying their lives. I’m supportive of entrepreneurs like sex workers making money on their own terms. When I was keto I couldn’t care less how many carbs other people ate, I always wanted them to enjoy their food the way they like. Not the way I like.
Not everyone you help in life has to be exactly like you. In fact, people just like you probably don’t need your help. The market is over saturated with productivity products. Why? Because entrepreneurs love productivity and build products for themselves. If you build for people dissimilar from you, you will find bigger market gaps. If you find these people fun or interesting, helping them will be an enjoyable time.
So I hope that explains a bit about my perspective on Wishtender and where my mind is at.
Twitter with Armando @Corlaez
Armando, who has been helping me, asked about WishTender:
I thought about it overnight before I got back to him.
Here are some things on my mind.
- One intangible problem I have I’m having trouble knowing how long things will take. For example, when I deployed, I deployed for the first time ever. So I had no clue how complicated cors would be and how much more complex a live environment is. I estimate when I’ll be done with a task but I’m always wrong.
- A more solvable problem I have is That I need to make sure the app is secure, I think I need to protect against csrf attacks. I just need to go through what else I need to do for security
- But I’m also just not done with a lot of the app yet. I’m still going through the backend validations. And the front end has some very messy unfinished pages.
- Amidst all of this people have been volunteering to pr, and I’m unsure how to go about communicating and managing them.
- I’m unsure of how minimal to make a beta product. I hear of mvps, but like I don’t know what can be left out. For example, security takes a while to implement, but it seems necessary even for an mvp.
-Dash, twitter DM to Armando
Armando said that this is just something you learn with time.
Adam Waselnuk @AWaselnuk also advised me:
Software time estimation is so hard I've personally written it off as an exercise in futility. You'll get slightly better with experience but it will always elude you. The reason is twofold:
- Software systems are complicated
- You are creating something new every single time you build. (usually)
If you created WishTender 10 times back to back, yeah you would get better at estimating. But would you create the same thing 10 times? No you would build dream phones and other marvellous wonders.
Just make sure you focus on important stuff every day and you'll ship soon!
-Adam, twitter DM
Armando brought it to my attention that this question is a fallacy. I assumed that mvp and beta were the same. 278 days into coding wishtender and I never considered that mvp and beta might be different ideas!
MVP and beta is not exactly the same to me. I would leave out a lot out of an MVP (most of the product and a lot of security) everything that is not core, every convenience would be out. Beta has more convenience and more security involved yes.
The idea behind MVP, in my experience, is that you want to build something that works and can be tested. Even if it is lacking features. And then grow it iteratively until you reach a beta status. However, every step of the way, the app should be working as expected.
...You do want to get to users soon, but landing a buggy app may cost you their trust.
Hence the point of having an mvp that you can validate it works* and then sprinkle some security a convinience feature or two and start getting user feedback
-Armando, twitter DM
In Armando's definition, an MVP is an app you test internally, while beta has added security and features so that it can be tested by actual users.
After this discussion, I found many definitions of MVP that differed from Armando's, but it was clear that beta was not the same as an MVP.
Armando...made me realize I’m not making an MVP: MVP and beta are not the same. Which is why I had some confusion about MVP’s. Armando described MVP’s as a insecure app you test internally, and beta as a secure app you have users test. Some others said you could also say an MVP is a mini product that tests a hypothesis- like drop box made a video describing the product they wanted to make and in response got a bunch of signups. I thought this was called a demo. But I guess a demo + feedback = mvp.
-From my journal 4/17/20
If demo + feedback = mvp, then I have already made MVP's.
If this is the case I already have done an mvp in part- I did market research and got signups. I have posted screenshots along the way and saw interest. My landing page describes the product as well, sort of a demo, and people were excited about the product. So maybe my business is more validated than I thought.
-From my journal 4/17/20
Armando said to focus on a stable app before approaching security.
All in all what I was trying to say is that IMO the page bugs should have a higher priority than security. I wouldn't bother with security and beta until I achieve some stability on the app
-Armando, twitter DM
Armando suggested using the issues tab on github to help communicate with volunteers.
I think a good idea to go about volunteers is to use the issues tab in github. There you can document what needs to be done and where, what you know and what you don't of each challenge you have. Volunteers can then skim through it and discuss the topic with you or others and eventually somebody will offer to solve that issue themselves. All that async documentation/communication is really useful IMO.
-Armando, twitter DM
I took Armondo's suggestion and I found the issues tab useful to keep track of what I need to do and what I've tried.
As I mentioned, estimating the time needed to complete each step of the app has been challenging. So, I'm not finishing in time for an April 20th launch.
I need to pick a new launch day. When I miss a launch day I get slow- not lazy- but I start focusing on details, the best way to code a feature, learning how things work. A slower, knowledge seeking, perfectionism mode. When I had a launch date I worked faster. I will pick another launch date, may 15. It is 28 days away. Last time I picked a date 30 days away. Hopefully this date is realistic. I didn’t account for the complexity of deploying the front and backend- cors, cookies behaving differently, browser differences.
I will focus first on stability, and worry about security after. A suggestion made by Armando Corleaz.
-From my journal 4/17/20
WishTender and Dream Phone Update (from journal)
4/5/21
This is a journal entry that I’m going to move to my log.
I haven’t updated my log. I’ve been very busy. So I’m going to take my journal time to give an update.
WishTender: Shakima and Adam, my friends on twitter, inspired me to pick a launch date. Shakima because she picked a launch date for her project, and Adam because he basically dared me to. I picked April 20th because it was 30 days away at the time. But it’s gone faster than I thought. I’ve been attacking the validations, and deploying the front and back end, and also working with Nat Sharpe who volunteered to help. Deploying is kind of complicated. There’s not as much information on it as there is other subjects. You encounter problems with CORS, browser defaults, cookies, ssl, DNS, and react build problems. Remote servers are so much more finicky than localhost. Having a launch date has made me reduce my decision making time. I don’t look for the best option to a problem. I just look for an option and go. April 20th is coming sooner than expected. So will I actually hit that date? I don’t know. This week the internet at my airbnb drastically slowed down and I will be traveling for 3 days this week. It might put a dent in my plans. But I’m going to try my best. I might be able to do it if I stop wasting time on small decisions. Either way, having this goal has helped shift my perspective which has been useful. Before I was focused on the details of my app. I would over estimate the importance of certain features like making the app work in every country possible. Now I’m more focused on quickly shipping because I created a deadline. So I prioritize the important features. I accept that the app can be improved and scaled after launch and it doesn’t have to be perfect. If it has to be scaled, that’s a good problem to have and I will find people willing to help. People will want to help if they know the app is valuable. I have been trying to ship an MVP but It’s taken me 8 months. I believe I created something bigger than an MVP unknowingly. It’s neither good or bad. I did more than I had to, but I learned a lot about software. The one thing I do want to make sure, is that the app is secure for my users to use. I wonder how do you balance that with fast shipping? Because you wouldn’t want to put your users at risk just you have to validate a business quickly. I’ll figure it out.
Btw, all of my WishTender code is public on github. You’ll have to search it because I’m on my phone journal.
Dream phone:
The dream phone is my device that allows me to communicate from a lucid dream. The Dream Phone SD version lets me save notes from a lucid dream to a text file. This might sound super weird and sci-fi but it’s totally a real thing and not really that complicated. I posted the code for my dream phone sd, but didn’t post a lot of explanation. I hope to in the future. It’s also on my github so you can search and find it.
The dream phone SD card version is done but I didn’t set it up yet or test it because I’ve been rushing to finish WishTender.
News fast: In order to help me focus, I’ve been on an intense news fast. I won't get in to all the details of why this is so great, (it's a lot), but it’s one of the best things I’ve done. I have a raised HRV (an indicator of low stress) which I believe is related. I love this article on quitting the news by David Cain https://www.raptitude.com/2016/12/five-things-you-notice-when-you-quit-the-news/
A lot of people say watch the news to stay informed but 'The news isn’t interested in creating an accurate sample. They select for what’s 1) unusual, 2) awful, and 3) probably going to be popular. So the idea that you can get a meaningful sense of the “state of the world” by watching the news is absurd.'
Dream Phone Update
Last time I posted about the dream phone, the device had a problem where it picked up electronic noise from the wall plug. This noise interfered with the electrode signals. Forum post.
I fixed this issue by switching from AC power to battery power.
I thought switching from AC to battery pack would require math and a bunch of resistors. But the LED strip has resistors built in. So it was a lot easier than I thought.
Building WishTender Update
After I posted about my challenging experience with the freelancer, everyone on Twitter pretty unanimously thought suggested that the freelancer was incorrect. Not only can react scale, but I shouldn't even care about scaling at this point.
The freelancer said he would still work on the react app with me, even thought I didn't want to take his suggestion of starting from scratch. But I told him I was going a different direction. I realized that, though he meant well, his goals were not aligned with mine and we wouldn't be a good fit. His personality was too pushy for me. He made assumptions without asking enough questions. This might work for some, but it's not optimal for collaborating with me.
There are an abundance of freelancers. So if I don't feel right with one, there are many many more.
Trust is important in collaborations.
There are two ways to work with someone you trust:
- Build trust up slowly with a new acquaintance
- Work with someone you already trust
If a new acquaintance loses your trust within the first few minutes, it's unlikely they will be able to grow it in the long run. In this situation, trust your gut!
Thanks to the many people who helped bring me clarity:
Namely:
- Matt Schiffman @darrownet
- Adam Waselnuk@AWaselnuk
Who spoke with me on zoom.
But also:
- Helge Drews @helgedrews
- Andrew @BrisaDotIO
- Khawar Jatoi @khawar_jatoi
- Marco @Wridgeu
- Shakima F. @Mostdefkima3ff
- Tejveer Singh@stevegremory
- Srđan Prodanović @some1else
- Marco Rivadeneyra @marcorivm
- Nat Sharpe @natsharpe
- Eric Howey @erchwy
- keyboard keelan @keyboardkeelan
- SpeakingSoftware Show / Philip @SpeakingSoftwa1
- William J. Ghelfi @trumbitta
- Brandy Wine @winebrandy48
- Sammy Rosen @sammyrose
Building WishTender Update
Today I'm feeling insecure and unsure. I thought I should post my thoughts to share the reality of my process. The following is the ramblings of highly confused person who just had a talk with a fiverr seller:
I just had gut wrenching meeting with a freelancer who suggested we start from scratch on Wishtender. He said there is going to be so many problems when I deploy Wishtender with react that it's better to just use django and html and css and start from scratch. And that django is more scalable. But I've been working on WishTender for like 7 months! So the idea of starting from scratch hurts my soul. I thought I could build a more complete app in order to learn about development while also aiming to launch a business.
Part of me thinks he's right. It's what everyone's been telling me all along. Go faster. Make an MVP. But I'm so confused, why does react even exist if it's neither good for mvp's nor scalable?
I can't even imagine that all that work I did with figuring out how to integrate with Stripe and calculating the fees and conversions, integrating with the exchange rate api, etc- that it will be easier from scratch. That's not part of the expertise of a Django developer.
I think I'm so stuck on the sunk costs though. This guy says it would take a month to complete from scratch in django. That's still faster than it might take to complete in MERN. But it feels like I would lose those 7 months of work. What was it all for. That's a sunk cost though. That's a bad reason to avoid switching. But there are other good reasons not to switch. 1. He doesn't know the whole picture of what I've done so far or what the project entails. 2. It's easy to create a react native app from react. 3. If I switched to django I feel I would need to rely fully on him then, since I don't know django. Am I just clenching on to the identity of a developer? I want to own a business. I don't even care about being a developer.
I don't know if I'm looking for reassurance that my way is right or looking for proof that I should change paths and go with the django mvp. But I should just be looking for truth. And the truth is probably gray.
Is this guy leading me astray or am I way off? Is it flawed to go with MERN? Am I overvaluing what I've done or is he underestimating what I've done? Is it worth making an MVP for an app that is almost finished? Is it worth putting $100-$1000 into an MVP for an idea that hasn't been validated?
My fear is that this suggestion to start from scratch- it maybe comes from a misunderstanding of what I've already done. And what goes into this project. My fear is that it will be a detour that will take longer. He didn't look at the entire app. He didn't even think I had an api. He had little information. But is it true that it will be difficult to deploy? Is it true that it will be difficult to work on with other people?
He said that it will be difficult to continue this project since I may have implemented things differently than him. But that confuses me. Isn't that just part of being a developer, you have to work with the code you're handed?
Maybe I'm seeking out the wrong person. Maybe he isn't an expert in taking over an existing project. He is an expert is starting projects from scratch, so he suggested starting the project from scratch. I need to calm down. Not over value one person's opinion. But don't under value it as well.
What is his expertise? He probably isn't used to taking over a project, pair programing. People come to him with blank canvases. So he wants a blank canvas. Where can I find people who work on canvases that have already been painted? Maybe if they suggested to start over I would feel more reassured.
What does a $1 MVP look like?
Is it possible to make an MVP for very cheap and continue working on Wishtender code?
Do I have to do a $1000 MVP? Is that worth it if I haven't validated?
Is a month + $1000 worth it? Is that really an MVP?
Is it still possible to move faster on WishTender with react?
I feel like my WishTender code is neither an MVP nor a perfect app. It's in a weird in between and it's causing me confusion.
I just need more clarity. This is just one guy. But is his point valid? Is it valid at all? It's true he doesn't have the full picture. But is there some truth to what he is saying? How can I find that truth?
I think I need to talk to some people and find out. Especially my mentor, because he's seen my code. I believe there may be truth to what he is saying. But I'm not sure the best route.
The more I think about it thought the more I think his advice was short sighted. He didn't look at my code. He didn't hear my requirements. I remember my mentor looked at my code, and was surprised that it was as good as it was. So maybe the fiverr guy doesn't know that my code is more complete than he would expect.
The fiverr guy also called me "sir" at one point in our chat before I went on video- which shows a pattern of miscalculation! If he thought I was a man, what other false assumptions is he making?
I'll have to think a bit more.
This whole thing is a distraction from what I was trying to do- just outsource a portion of the app. Should I scratch that idea? Should I continue regardless?
Building WishTender Update
- Hiring Freelancers
- My questions about assigning tasks:
- isolating?
- creative licence?
- communication?
- My questions about assigning tasks:
- Simulating Working With A Freelancer
- Mistakes I Discovered when Assigning Tasks
- Resources & Notes
- Takeaways
I want to experiment with outsourcing parts of WishTender. But I have no clue what it looks like to work on a team.
I'm wondering:
- How do I isolate tasks for freelancers or team members into manageable tasks?
- How much creative licence do I give freelancers? Do I give them a highly specific tasks that define not only what should be done but how? Or do I give them general goals where they can take some creative licence in UI, UX, and code design?
- How do I communicate the task to them effectively? What do I need to provide them with in order to accomplish that task?
To prepare for outsourcing I've been practicing working as my own employee using gitflow and github.
This gave me some insight on how assigning tasks can go wrong.
I had a bug in my app. I thought it orginated in the frontend. So I assigned it to FakeEmployee, my alter ego. FakeEmployee is a frontend dev. But FakeEmployee realized the bug was in the backend. So fake employee had to go back to me and tell me to fix the bug. How do I outsource a bug if I don't know where it originated and who is capable of fixing it?
I assigned FakeEmployee a simple task; create tabs in my menu with material UI that underline when on the respective page.
I created a design on Photoshop to give to FakeEmployee.
I thought this would be an easy task for FakeEmployee. But it was more complex and involving restructuring a my app. FakeEmployee had to post on stackoverflow for help. And the issue was even too complex for anyone there to solve. So while FakeEmployee got help from stackoverflow, ultimately FakeEmployee had to post the correct solution herself.
If I was working with a real freelancer, I might have underestimated the time and effort of the task. So I might have offered too little money.
In addition to paying my freelancer appropriately, if I knew the complexity of the task, maybe I would have decided against the highlighting tabs. Perhaps the highlighting of the tabs is a diminishing return for the effort needed to implement them at this stage. I might have gone with a simpler task, like tabs that don't highlight.
I wanted to get a better idea of how to manage freelancers and a team. Here's some of the resources I consumed.
This video explained what a DevOps team is, what problem it solves, and what building software in devops looks like.
- Open API
This really helped me understand what teamwork looks like working with a product manager, designer, and frontend dev. Only I didn't see how the backend developers fit in.
This video by TechLead @techleadhd showed how workflows between team members can flow multiple ways. For example, instead of always following the product manager's lead, programmers can consult the pm on the best approach. He also spoke about consistency in design and holistic design. Something I half implemented with Material UI themes. Sometimes I use Material UI theming, sometimes I when did not. I may need to work on keeping my UI consistent. When I make a Photoshop mockups, I may need to refer to Material UI docs to make sure I'm making something that can be implemented using the theme. I can also design a mockup and ask for feedback from the developer. Maybe they will notice that the design unnecessarily veers off from the theme and can make suggestions to improve the mockups.
How does a front-end developer work with a back-end developer?
Answer:
For me, the best approach is to:
Have the front-end developers provide a “wish list” of APIs they want - and provide documentation of the desired behaviors.
Have the back-end developer review the desired APIs. If some appear to be difficult to implement or are likely to be slow, the back-end dev should see if other alternatives are acceptable to the front-end dev.
Both teams implement their code to the agreed APIs. (Note: I sometimes have the back-end team implement “stub” versions of the APIs, that return test data to allow the the actual interaction to be tested before the full implementation is complete.
** ITERATE AS REQUIRED TO MAKE IT BETTER** - this is the most important step. Occasionally, the teams will be happy with the initial designs and implementations. Most often, one or both of the sides of the APIs will want changes - plan on a few revisions after the initial implementation is finished.
Note: this approach means that developers will be asked to change code that works and meets the (original) requirements. This is something that developers need to learn to accept. I generally assume that people requesting a change to requirements have a good reason for the request. Much of this arises from the fact that in implementing and/or using an API, developers learn a lot they didn’t know when coming up with the initial requirements.
Video: How To Create Your First Wireframe
We use API description langauges in order to define a contract.
Contract before code ...but scenarios, use cases and other API design work still come first.
In Open API and JSON schema the contract is an open world system.
Open World System: You define what is possible but we don't tell you everything that is potential possible.
Example your API description could leave out 500 responses.
Resources can be grouped by tags.
You don't need to include everything in your contract descriptions. You have other places to document. Changing stuff in your contract can create breaking changes.
There are tools that allow you to use mock servers with your API contract so client developers don't have to wait for a server implementation to complete.
My question: Is there a way to generate React routes from a contract?
Documenting. Documenting the API, tasks that are complete or need to be completed, using git to document changes. All of this helps with communication.
Cross-collaboration and bi-directional workflow: Different roles (designers, developers, operations) can communicate constraints within the system to help everything come together. There can be a lot of back and forth between these roles instead of just a top down flow.
Building WishTender Update - Tuesday 3/16/21
To set your email and username as the author.
$ git config --global user.email "secondaccount@gmail.com"
$ git config --global user.name "SecondAccount"To show the config lists
# global settings
$ git config --list --global
# local settings
$ git config --listTo delete some property from the config. There is no such configuration as user.username so I deleted it after accidentally adding it.
# delete user.username from global settings
$ git config --unset --global user.usernameIf you change the configurations, the author will be set to the new account. But you will still be pushing to github through the first github account. So if your want to also switch git hub accounts:
$ git credential-osxkeychain erase
host=github.com
protocol=https
[Press Return]
The terminal is going to ask for new credentials when you push.
If that doesn't work also try:
git config credential.username 'SecondAccount'
Building WishTender Update - Saturday 3/13/21
- Why I'm making my repo public
- Security by obscurity
- Build In Public
- What About Competition?
- Loss Aversion & Conclusion
Features on Github that make collaboration easy are limited to paid accounts. You can get around paying, if you keep your repository public.
WishTender is a private repo. Now that I'm looking to work with other devs, I need access to these colab features. So I decided to revisit opening the repo.
I originally made my WishTender repository private because I was afraid of insecurity. "The more people know about your project's backend, the easier people can hack into it." I struggled with.
It's really important for me to participate in #BuildInPublic. Both selfishly and altruistically. #BuildInPublic is my way of giving value to the community at every point of my journey. Sometimes it can feel like you have nothing to give, but you can always share you're journey and what you're building. That is the altruistic side of #BuildInPublic. Selfishly, it's also given me many opportunities I wouldn't have had if I built and learned in private.
I posted my dilemma back in September. Something Brennan Hitchcock @k0ode said always stuck with me:
Security by obscurity should never be your primary line of defense.
Brennan continued:
Furthermore by having security by obscurity included in your cyber-defense plan it becomes a crutch and a blind spot.
Brennan encouraged me to "Build in public!"
Even though he made a good point, I figured I was too novice to rely on anything but security by obscurity. So I continued to keep my repo private. I've learned a lot since September and now have a fuller understanding of what security involves. I feel more confident that I can post my code and build strong security before I launch.
I also found this question and answer on Quora useful:
Quora Answer by Aravind Ravi-Sulekha:
Generally, no.
The only exception to this rule is if a large part of the value of your startup comes from some proprietary or otherwise non-obvious algorithm expressed in your source code. For the vast majority of technology products, this is not the case — value usually comes from a clever user interface or other features that are straightforward to implement once it’s seen in action.
If a competitor with their own product wants to copy a feature you invented, it is almost certainly easier and cheaper for them to implement it independently than to figure out how to adapt your code to their architecture. Making software components work together when they weren’t designed to is extremely hard.
That leaves competitors willing to build their startup entirely on your code. It isn’t clear why someone would want to do that — they can’t compete with you by doing exactly what you’re doing. You start with an enormous advantage, having the team which wrote the original code and therefore the ability to move much faster. As long as you don’t accept outside contributions to your project, you can stop publishing code changes at any time.
My startup, Scrollback, uses an open GitHub repository for almost all its code. The only exception is a plugin that implements a learning algorithm for organizing threads of discussion — something which is not straightforward to implement even after you’ve seen it work.
I'm going to take what feels like a more courageous and abundant mindset and make my repo open.
Keeping it closed was based on my fear of loss. Not hope of gain. I was exhibiting loss aversion. Loss aversion is the tendency to prefer avoiding loss to acquiring equivalent gains.
I fear being hacked, I fear someone coming in and copying everything WishTender is doing.
But what if instead I focused on what I gain by making it open? I'll help others who are looking to build. Maybe I'll get more help from experienced onlookers. Maybe it will bring more attention to my work and WishTender, bringing me more opportunities.
Because of this, I will be making my repo public. What do you all think? What are the positives and negatives of making a startup repo public? Is it something that should only be temporary? Am I making a huge mistake? Let me know @DashBarkHuss
Building WishTender Update - Tuesday 3/3/21
Todays Post:
I'm going to take a big leap and outsource some parts of my project. It's very scary and exciting. I've never coded with anyone. But I've always felt like a real entrepreneur should know how to outsource. And reading the book The Entrepreneur Roller Coaster by Darren Hardy is giving me extra inspiration. He highly stresses outsourcing. "Be a quitter"- quit as many parts of your business as you can. Get people better than you to do those tasks.
I want to see what outsourcing some bits of my code will do.
To prepare, I created a WishTender Organization on github and created a github account called FakeEmployee. I am acting as my own employee, making commits to my repo under the FakeEmployee account.
I also did som micro outsourcing. I hired a machine learning guy on Fiverr. $5 to talk about the feasibility of a project. The project had nothing to do with WishTender- it was an app for lucid dream research on behalf of Karen Konkoly & Northwestern. I don't know if I will pursue the project, but I wanted to try out Fiverr anyways. It was a useful introduction to outsourcing.
I'm kind of surprised how much access Github gives this FakeEmployee account. They can just push straight to master if they want. I believe you can prevent this by creating a branch protection rule for the master branch. But it's not free unless I make my repo public:
Protected branches are available in public repositories with GitHub Free and GitHub Free for organizations, and in public and private repositories with GitHub Pro..."
So I either pay for github, make my repo public, or trust that freelancers won't push to master.
Does anyone know if I'm correct about this? Do I have to pay or make my repo public? Which would you do?
article: use github professionally
video: use github professionally
video: GIT Workflow - Georgia Tech - Software Development Process
The gitflow workflow - in less than 5 mins.
Building WishTender Update - Tuesday 3/2/21
Todays Post:
I posted on Twitter:
I got a response from "Nomad"
"Nomad" offered to zoom call with me.
"Nomad" was a really cool. His real name is Matt Schiffman. He's a self taught programmer with an impressive resume, an entrepreneur, and a fellow Arduino enthusiast. A great teacher too.
I'm grateful that I'm able to shout "Help me!" into the world and get a response. This is the useful side of social media- real community.
I wondered why he was so eager to help? It turns out he had an interest in my dream phone because one of his friends was paralyzed in childhood. The same technology used for the dream phone, could be used to help paralyzed patients communicate.
I almost didn't post a recent update of my dream phone. Part of me thought, "No one cares- it's not finished!" But because I did share, Matt knew what I was up to and wanted to support my growth. The lesson: don't shy away from self promotion. People want to help you and you can help them. This is how community works.
As a self taught solo developer, I work on the frontend and backend at the same time. When I'm working on the frontend, I might realize that it would be helpful if certain data was available to this view. Then I'll go in the API add the data. Then continue working on the front end. I do this a lot!
Clearly, this won't work when working with a team. I can't, at every twist and turn, assign new micro tasks to the front and back end devs.
Matt said that you first start with what data you want. What data do you want the API to feed to the front end?
If you think in terms of the data- the meeting point of the front and back end- you can then build outwards the backend and frontend.
You can create fixtures (aka dumb data, contracts). A fixture is sort of like a stub, or mock, in testing. But it's to mock up what the API will send to the front end. I'm guessing it can also be what your frontend will send to your backend.
This is what branching might look like when working with people:
Production->Dev->feature->WB1(Working Branch)-&-WB2
Your contractors/employees will create pull request. Which you can review and decide to merge or not.
Learn more about this type of flow by checking out Gitflow:
Gitflow Workflow is a Git workflow that helps with continuous software development and implementing DevOps practices. It was first published and made popular by Vincent Driessen at nvie. The Gitflow Workflow defines a strict branching model designed around the project release. This provides a robust framework for managing larger projects.
It's important to learn project management as an entrepreneur. You will eventually progress to less coding and more managing.
OmniPlan gives you the power to manage complex projects, coordinate teams, and continuously improve your processes.
They have free tutorials on project planning. You can get a 2-week trial to try out Omniplan. Using Omniplan is a good way to learn about project planning. Afterwards, you can apply what you learned to general project planning, without strictly using Omniplan.
MongoDB Compass
Compass is a GUI for working with MongoDB.
The most innovative cloud database service on the market, with unmatched data distribution and mobility across AWS, Azure, and Google Cloud, built-in automation for resource and workload optimization, and so much more.
For storing user content/images.
Amazon S3
Object storage built to store and retrieve any amount of data from anywhere
Building WishTender Update - Tuesday/Wednesday 2/24/21
Todays Post:
I'm saving some information here about Stripe cross border payouts and default currencies.
Cross-border payouts enable platforms in the U.S. to pay out to connected accounts in the following countries Source
Note: In the post below I did a lot of unnecessary work to find what looks more easily available in Country Specs, a Stripe resource.
Supported Countries (non-US):
const supportedPayoutCountries = [
{ name: "Australia", code: "AU" },
{ name: "Austria", code: "AT" },
{ name: "Belgium", code: "BE" },
{ name: "Bulgaria", code: "BG" },
{ name: "Canada", code: "CA" },
{ name: "Cyprus", code: "CY" },
{ name: "Czech Republic", code: "CZ" },
{ name: "Denmark", code: "DK" },
{ name: "Estonia", code: "EE" },
{ name: "Finland", code: "FI" },
{ name: "France", code: "FR" },
{ name: "Germany", code: "DE" },
{ name: "Greece", code: "GR" },
{ name: "Hong Kong", code: "HK" },
{ name: "Ireland", code: "IE" },
{ name: "Italy", code: "IT" },
{ name: "Latvia", code: "LV" },
{ name: "Lithuania", code: "LT" },
{ name: "Luxembourg", code: "LU" },
{ name: "Malta", code: "MT" },
{ name: "Netherlands", code: "NL" },
{ name: "New Zealand", code: "NZ" },
{ name: "Norway", code: "NO" },
{ name: "Poland", code: "PL" },
{ name: "Portugal", code: "PT" },
{ name: "Romania", code: "RO" },
{ name: "Singapore", code: "SG" },
{ name: "Slovakia", code: "SK" },
{ name: "Slovenia", code: "SI" },
{ name: "Spain", code: "ES" },
{ name: "Sweden", code: "SE" },
{ name: "Switzerland", code: "CH" },
{ name: "United Kingdom", code: "GB" },
];To get this list I put the below code in the devtools console of this stripe page in order to get the countries:
arr = [];
for (i = 1; i < 34; i++) {
arr.push(
document.querySelector(
"#content > section.availability > ul > li:nth-child(" +
i +
") > span:nth-child(2)"
).innerText
);
}Then to get the country codes (of non-US countries) from the names, I had to use the country-data library;
const countryData = require("country-data");
const countries = [
"Australia",
"Austria",
"Belgium",
"Bulgaria",
"Canada",
"Cyprus",
"Czech Republic",
"Denmark",
"Estonia",
"Finland",
"France",
"Germany",
"Greece",
"Hong Kong",
"Ireland",
"Italy",
"Latvia",
"Lithuania",
"Luxembourg",
"Malta",
"Netherlands",
"New Zealand",
"Norway",
"Poland",
"Portugal",
"Romania",
"Singapore",
"Slovakia",
"Slovenia",
"Spain",
"Sweden",
"Switzerland",
"United Kingdom",
];
countries.forEach((coun) => {
const code = countryData.countries.all.find((c) => c.name == coun).alpha2;
const obj = { name: coun, code: code };
console.log(obj);
});I ran a script to get the default currencies (of the supported non-US countries).
I later found out about a stripe resource that may have helped me: Country Specs. Since I didn't no about country specs, I did extra work to get the default currency- creating stripe accounts for each country and logging the default currency.
Here is that script I used to get the default currencies. Using Country Specs would have been more efficient:
const stripe = require("stripe")(process.env.STRIPE_SECRET_TEST_KEY);
const createAccountInfo = (country) => {
const info = {
country,
type: "express",
//
capabilities: {
transfers: {
requested: true,
},
},
};
if (country !== "US") {
info.tos_acceptance = {
service_agreement: "recipient",
};
}
return info;
};
const delay = (millis) =>
new Promise((resolve, reject) => {
setTimeout(() => resolve(), millis);
});
const createAccount = async (country) => {
const info = createAccountInfo(country);
try {
const account = await stripe.accounts.create(info);
const object = {};
object[country] = account.default_currency;
console.log(object);
await stripe.accounts.del(account.id);
} catch (error) {
console.log(error.message, " ", country);
}
};
const supportedPayoutCountries = [
"AU",
"AT",
"BE",
"BG",
"CA",
"CY",
"CZ",
"DK",
"EE",
"FI",
"FR",
"DE",
"GR",
"HK",
"IE",
"IT",
"LV",
"LT",
"LU",
"MT",
"NL",
"NZ",
"NO",
"PL",
"PT",
"RO",
"SG",
"SK",
"SI",
"ES",
"SE",
"CH",
"GB",
];
supportedPayoutCountries.forEach((country, i) => {
setTimeout(() => {
createAccount(country);
}, i * 1000);
});//results of the script turned into an array for convenience. Non-US countries, remember to add US
const defaultCurrencies = [
{
AU: "aud",
},
{
AT: "eur",
},
{
BE: "eur",
},
{
BG: "bgn",
},
{
CA: "cad",
},
{
CY: "eur",
},
{
CZ: "eur",
},
{
DK: "dkk",
},
{
EE: "eur",
},
{
FI: "eur",
},
{
FR: "eur",
},
{
DE: "eur",
},
{
GR: "eur",
},
{
HK: "hkd",
},
{
IE: "eur",
},
{
IT: "eur",
},
{
LV: "eur",
},
{
LT: "eur",
},
{
LU: "eur",
},
{
MT: "eur",
},
{
NL: "eur",
},
{
NZ: "nzd",
},
{
NO: "nok",
},
{
PL: "pln",
},
{
PT: "eur",
},
{
RO: "ron",
},
{
SG: "sgd",
},
{
SK: "eur",
},
{
SI: "eur",
},
{
ES: "eur",
},
{
SE: "sek",
},
{
CH: "eur",
},
{
GB: "gbp",
},
// US
// {
// US: "usd",
// },
];I asked on StackExchange:
Email Confirmation links must be GET, but not safe:
When a user signs up to a web app, they often get a confirmation email. This email will contain a link. Once the user clicks the link, the app confirms the user's account as correctly associated with said email, perhaps by changing a
confirmedfield on theuserdocument fromfalsetotrue.My question:
- The endpoint changes data (ei:
user.confirmedfromfalsetotrue) so this action isn't safe*.- The link is a GET request.
Don't these two facts violate HTTP standards? GET requests are supposed to be safe. So
PATCHseems more appropriate for changing theconfirmedfield. However, there is no way to make a link in an email aPATCHrequest.How do software engineers handle this inconsistency? Is it just a violation that's accepted? Or is there a correct way to implement this so that we don't violate HTTP standards.
*An HTTP method is safe if it doesn't alter the state of the server. source
From the comments it seemed like a lot of people never considered this issue.
Some people disagreed on whether is should be considered safe or not to change the confirmed field. They said it's safe if the user isn't knowingly making this change.
These arguments all seemed "lawyer-y"- not getting to heart of the problem. People just focused on the semantics in question, and disregarded why these standard exist. You can change the definitions of words all you want but the software doesn't care. I'm not against breaking rules, but it seemed many lacked understanding of the purpose of the rule. You should understand the rule before you consciously decide to go against it.
The answer I selected focused on the why- why it's recommended to only use safe actions with GET. And then gave a solution that addressed these concerns.
This next question I asked on Stackoverflow. There were no answers. I isolated the problem into a small bit of code and put it in a repo so you can easily recreate the problem.
Mongoose expiration not setting- expiration not showing in getIndexes:
I have a document that has an expiration set for 2 minutes.
const docSchema = new mongoose.Schema({ createdAt: { type: Date, default: Date.now }, expireAt: { type: Date, default: Date.now, index: { expires: "2m" } }, }); const Doc = mongoose.model("Doc", docSchema);When a
Docis created (ie.const doc = await Doc.create({});), I would expect it to have the expiration set. However, sometimes it's setting- but sometimes it's not.I check if there's an expiration on the
Docby runningdb.docs.getIndexes()in a Mongo shell. The result will either show the expiration information:[ ...{// some other object}, // expiration info---> { "v" : 2, "key" : { "expireAt" : 1 }, "name" : "expireAt_1", "ns" : "expireDocument.docs", "expireAfterSeconds" : 120, "background" : true } ]Or it will leave it out. I'm guessing it leaves it out when it didn't apply the expiration to the documents.
It seems sort of random when it gets set and when it doesn't. However, I was able to figure out a consistent situation where it is set and where it isn't so the problem can be recreated. Though, I have no clue what separates these situations besides the order they are in.
Download the small repo here. You can also see the
server.jscode below.Open the folder in VScode.
CTR+` to open terminal. Run
npm iClick the debugger icon. At the top left play
Expire Documentdebugger.After the script creates a
Docthe debugger pauses online 15. While paused, open a new terminal.Run
mongoand thenuse expireDocumentto navigate to the correct database.Run
db.docs.getIndexes()to see the indexes. You will see the second object contains info about the expire information for `docs as expected.[ { "v": 2, "key": { "_id": 1 }, "name": "_id_", "ns": "expireDocument.docs" }, // second object--> { "v": 2, "key": { "expireAt": 1 }, "name": "expireAt_1", "ns": "expireDocument.docs", "expireAfterSeconds": 120, "background": true } ]
- run
db.docs.drop()to drop the collection- Unpause the debugger.
- Again, after the script creates another
Doc, the debugger will pause online 22. While it's paused rundb.docs.getIndexes()again. For some reason, this time there won't be any second object that indicates there is expireAt information.[ { "v": 2, "key": { "_id": 1 }, "name": "_id_", "ns": "expireDocument.docs" } ]Why am I not getting consistent expirations set?
const mongoose = require("mongoose"); const docSchema = new mongoose.Schema({ createdAt: { type: Date, default: Date.now }, expireAt: { type: Date, default: Date.now, index: { expires: "2m" } }, }); const Doc = mongoose.model("Doc", docSchema); const run = async function () { // make document const doc1 = await Doc.create({}); debugger; // While paused: // 1. run db.docs.getIndexes() in mongo shell and there will be a second object with "key":{ "expireAt":1} // 2. then run db.docs.drop() to drop the documents // 3. unpause script // make 2nd document const doc2 = await Doc.create({}); debugger; // run db.docs.getIndexes() in mongo shell and if you dropped in step #2 when script was pause at line 14, there will be no second object with "key":{ "expireAt":1} }; mongoose .connect("mongodb://localhost/expireDocument", { useNewUrlParser: true, useUnifiedTopology: true, useFindAndModify: true, }) .then(() => console.log("Successfully connect to MongoDB.\n")) .catch((err) => console.error("Connection error", err)); run();
Notes:
From: How to remove TTL form MongoDB collection?:
Use
db.adInfos.getIndexes()to see the indexes on the collection.
I mostly have been posting about WishTender on here. But, in addition to WishTender, I've also been spending my time brushing up on my electronics in order to finish my dream lab.
From Log Entry on 04/05/20
I'm currently building a lucid dream lab the[that] will induce lucid dreams and enable a user to communicate from a lucid dream to the outside world. I started this project in 2012. But it's been an on and off adventure.
After participating in a lucid dream study at Northwestern, I was inspired to revisit the project and add some features that Northwestern had.
My goal is to Tweet from a dream by connecting the dream phone to the Twitter API.
Here's more explanation on how the device works, and how I got it to send a text (while awake but can theoretically also work while asleep).
The latest problem I had with the dream phone was that the device has some electronic noise from the wall plug that interfered with the electrode signals. Forum post here
Since adding a wall plug adapter,
I now get a fuzzy eye movement signals from the electrodes
instead of a clear readable eye movement signals.
The reason I added the AC power adapter was in order to power the red led strip. The led strip provides the light needed to induce lucid dreams.
I think the best solution is to switch from a 12V AC adapter to a battery pack to both reduce 1. the noise and 2. the danger of physically being hooked up to AC power.
But I'm not totally sure. So I've been brushing up on electronics basics- current, volts, resistance, Ohms law, Kerckhoff's voltage and current laws, working with components, etc.
Now that I've filled you in, I just wanted to use this space to save some notes and resources.
-
How Led strips are wired- go to last section.
-
Engineering Circuit Analysis- I bought the series but the first few videos are available on Youtube. Those first few videos were actually my favorite videos in the series. Most of the later videos I paid for just went over problems and solutions.
-
Tinkercad
Tinkercad has a virtual circuit maker. It allows you to create virtual circuits. You can use virtual power supplies, virtual components, and even test it all with a virtual multimeter. Great for practicing basic electronics without expensive components.
Building WishTender Update - Thursday 2/11/21
It's been 213 days since I started coding WishTender, my unfinished bootstrapped product. I've learned a lot about building an application.
Now, I'm starting to consider something both intimidating and exciting: coding with other people.
I've always envisioned myself transitioning out of coding WishTender, as lots of founders do. It would be powerful to learn how to delegate tasks and manage people.
First, I would like to start small. I'd like to get experience delegating some of the coding to other developers.
But, as a newly self taught developer who has never worked with other developers, I have a lot of questions and gaps in my knowledge.
-
Logistics: What does working with other people look like?
-
How do I go about using git?
I've only ever used git as a single person. If I could see what it looks like for people to collaborate on github that would be helpful.
-
How do I separate out tasks?
As a single developer I can work on the front and back end at the same time. I often approach tasks without separating these out. For example, if I work on the checkout front end, I will also be messing with the checkout backend at the same time; making tweaks to the API as I figure out what I need from the backend. I would imagine that I would need to separate tasks up when I work with others. Collaboration probably requires more planning, like having the backend already figured out, knowing which API calls will take place on each page, and knowing what data will need to come in and out of the API before telling someone to build a front end page.
Kenneth Wade, DevOps manger and founder of Kenjyco, suggested a good first step: try coding my app on two different computers. It will help me get an idea of how to use git as a collaborator.
-
-
Security: How do I keep from exposing too much information about the app with freelancers/new employees who I know little about?
- I wouldn't want just anyone to have access to api keys, customers info, or passwords to relavent accounts- like my stripe platform account. Yet it feels like sometimes I will need to give collaborators access to these.
-
Money: How do I afford to work with others as a bootstrapped business?
- I probably won't be taking on employees right away. But what about freelancers or interns? Or starting out by finding a friend or mentor to pair program on my app? When do I transition to hiring fulltime employees and what does that transition look like?
If you have any suggestions about how I can learn more about outsourcing, managing, collaborating let me know on Twitter or Linked in!
Here's a video about bootstrapping as a developer by Tekin Süleyman @tekin.
It's rare to find developers who are bootstrapping a product. More commonly, founders are non-devs raising funds or trying to find a technical co-founder. Sometimes in combination with using no-code tools to prototype. But being a developer gives me both unique advantages and disadvantages to consider. And certainly different goals from vc raisers.
Even in Tekin's talk, one of the first things he mentions is that bootstrappers should have one goal: to turn a profit. While conversely, funded ventures are going to aim for growth. A bootstrapped app would not want 100k users on launch day. They wouldn't have the means to support this traffic. Bootstrappers want steady growth that their profit can support. So my goal is profit. But funded businesses have or will try to get money to support growth, so their goal is growth. Growth shows investors that the business is viable. This is just one of the many ways these two frameworks might think differently.
It was also nice to see Tekin talk about his slow and steady growth. His first commit was June 2009. He launched in May 2012. A month later he had his first paying customer. VC funded projects might move a lot faster. I thought I was moving at a snails pace compared to what I saw. For example and speaking of fast, I think I saw the founder of Fast talk about how Fast was just an idea a year ago. I've been working daily on WishTender since June*. I am no where near the pace that Fast moved. But my journey is going to look different than a funded venture.
Exposing myself to different bootstrappers journeys will be beneficial.
Tekin goes into his journey as a bootstrapper and how and why his business failed and it's a great watch.
*Possibly an exaggeration. I believe I took off on Yom Kipper, and maybe some other special occasions or at least decreased the amount of work I did.
Building WishTender Update - Sunday 2/7/21
It's been 7 months and 2 weeks since WishTender was an idea. 6 months and 4 weeks since I made a demo of the app. 6 months and 3 weeks since I started coding the app.
When I started my spouse asked me when I'd be done. I said by the end of October. That was 3 months ago.
I'm not surprised it's taking me longer. I'm building a full app myself for the first time. I have to remind myself, it's only been 768 days since I started my first day of #100DaysOfCode.
So I'm not here to complain about the length of time it's taking me.
I'm at a point though where the excitement of starting the project is wearing off. This is the point I believe where most people struggle. The boring part where you're pretty far from the newness, and pretty far from the home stretch. It's going to take patience and persistance to get though this.
But I can't just rely on "persistence". I've found that forceful modes like persistence and discipline don't work as well as intrinsic motivation, even for people like me who have seemingly endless amounts of willpower. We can try to "discipline" ourselves to the finish line but we often pay the price- usually our health will suffer from the hormonal effects of extreme discipline.
So I need a tactic- beyond willpower- to get through this hump. What got me excited in the first place? That excitement is what I need back.
Seeing myself as the founder of a profitable company.
Seeing myself making money.
Seeing myself making enough money to fund lucid dream research and technology.
I shouldn't be ashamed to get too grand with it- Why not see myself on the cover of Wired Magazine? Meet Dashiell Bark-Huss: The Entreprenuer Who Tweets From Her Dreams And Put $1 Billion Into Lucid Dream Research
These are the grand delusions that I normally try to tone down. After all, it's more important to get to the blood and sweat- the coding the, market research, the work- than get swept away with day dreams. Or so I thought. I'm seeing instead, that we need a balance of the grand delusions and the hard work. Contrary to what "Discipline Dashie" believes, "Delusion Dashie" runs the show. My grand delusions are magic fuel.
What do I mean these delusions are magic fuel? I've been experimenting. I've found that after I visualize or journal how exciting the future might be, it feels like I instantly gain 15 IQ points and drank a cup of coffee. Much of my very real lethargy and difficulties go away completely after I rejuvenate my excitement for the future.
What are the visualizations I'm doing? Well, I haven't been so prescriptive with it. It's just something I've been dabbling with. For example, I tried the written equivalent of creating a vision board: Journaling why I'm building WishTender, what it will do for my life and other people. I also collected some pictures on instagram of others in similar positions to where I want to be. I updated my linked in with "Technical Founder at WishTender" to remind myself this is real and give myself a bit of public accountability.
Now that I'm seeing the necessity and benefits of visualizations, I may be a bit more prescriptive. So many self help book say to do some form of visualizations or goal reciting before work (The One Thing, The Miracle Morning, Think And Grow Rich). When you get far enough into a project this becomes so important. You have to keep that intrinsic motivation going. Other people aren't going to do be your cheerleader. Most probably don't have the patience for their own projects, they certainly won't have the patience for yours. So be unique. Be persistent, be patience, and have grand delusions.
Building WishTender Journey - Friday 1/29/21
Resources to save: Node Checkout example
Building WishTender Journey
Answers to My App Architecture Question
After my question yesterday I got a lot of helpful responses.
Theo Khayat on LinkedIn
You're looking for some all-or-nothing processing(checkout AWS step functions - am not recommending specifically AWS, just as an example of chaining multiple backend events) as this allows your frontend to request a single endpoint, and subsequently know if ALL the backend events succeeded or not, in a single status code. Hope that helps and best wishes :)
Theo introduced me to the term all-or-nothing processing. I googled all-or-nothing processing which lead me additional useful terms like batch and bulk requests. It lead me to better worded questions than my own like "How do I perform different actions on resources of different types in one request?"
Sometimes when I'm stuck it's because I don't know the correct terms or questions to ask. Reaching out for help can introduced me to terms and well-worded questions that will get me on track.
Ryan Lynch @joyofui on Twitter
My answer to your question, which I'd sum up as "What do I do when an endpoint doesn't map to REST", would be : not everything has to be RESTful in your API. REST and is just one way of slicing things up. Maybe think of some endpoints as REST, some as just RPC's like "doCheckout"
Me:
Yes, you phrased my question better than I did. A lot of my questions lately have surrounded this idea- that you do not alway have to follow REST. As a newbie it's been a bit confusing trying to understand when to break a convention. Thanks for bringing RCP to attention.
I should say when and how to break convention
[This part was in a DM in a related conversation:] When I was more of a newbie there usually was one best way. But it seems as you progress to more advanced things there are less best practices. But it's interesting to see how other developers think through the problem
A rule of thumb is, if you are creating either more work or more mental overhead to build or understand the code, think about changing conventions or breaking them all together. Not all problems are generic and have generic solutions, so don't be bound to generic answers!
This last remark by Ryan hits on a problem I see with myself and other newbies- fear of breaking convention. Often times it isn't even clear to a newbie when something is a flexible convention, and when it is a serious rule that you should never break.
I looked into other forms of API's after Ryan suggested RPC.
Video: Nate Barbettini – API Throwdown: RPC vs REST vs GraphQL, Iterate 2018
In this video Nate illustrates why you might break a convention for another API convention that fits your use case. It may be more important for one project that the API is highly flexible and documented, REST is great here. But I'm building a private API so I may value low chattiness over flexibility.
Helge Drews @helgedrews on Twitter:
I think what you're searching for is a CheckoutController and not CheckoutService. Simply said, you inject all your Services into it and the Controller orchestrates all the business logic, handles Errors and so on. Put it behind the POST /checkout route. :)
My Response:
Thanks helge, I had a checkout route (the equivalent of a controller?) but it was getting big and I though it was considered a best practice to leave fat business logic out of the controller and move it to a service, so I also made a service. Not sure if that makes sense
Before I sent that last response, Helge sent me a DM to elaborate:
Here I've got some more infos to the answer in your twitter thread about how to organize your app architecture.
in the root file app.js you can do something like this:
const checkoutController = new >CheckoutController({ emailService, orderRepository, paypalService, productRepository, shopConfigRepository, shoppingCartRepository, stripeService }); // ... app.use('/checkout/', >checkoutController.router); CheckoutController looks like this: export default class >CheckoutController { constructor({ emailService, orderRepository, paypalService, productRepository, shopConfigRepository, shoppingCartRepository, stripeService }) { // Routes this.router = new Router(); http://this.router.post('/prepare', async (req, res, next) => this.prepare(req, res, next)); http://this.router.post('/complete', async (req, res, next) => this.complete(req, res, next)); // Repo this.orderRepository = orderRepository; this.productRepository = productRepository; this.shoppingCartRepository = shoppingCartRepository; this.shopConfigRepository = shopConfigRepository; // Service this.emailService = emailService; this.paypalService = paypalService; this.stripeService = stripeService; } // the implementation of the methods >behind the routes and a lot of other >helper methods
Helge used the terms "repository" and "controller" which introduced me to the Repository-Service pattern.
John Zhao on LinkedIn:
one endpoint can interact with many services and models. that's the entire point of the backend - to glue all the different pieces together.
the vast majority of backend code in the real world is never as simple as one route -> one service -> one model. taking your example of getting a product from a database, in the real world, you'll probably want to call one endpoint to get the product info, product upsells and product reviews that are stored in different models.
just google node ecommerce projects on github and you'll find plenty of examples to guide you
Many tutorials simplify projects so we rarely see these more complex real world backends. Finding more complex examples is important.
It seems obvious, but I hadn't thought to search for the term "ecommerce", I kept searching for "shop". That small suggestion was helpful.
As I was reading guidance from the helpful people above, I did some googling and note taking:
What if I want to do DELETE /resources/ID1 and PATCH /another-resources/ID2 at the same time?
This is really nasty and definitely not REST, but it can be useful for backend for frontend or experience API for example.
To do that we’ll need to POST data on a specific endpoint which could something like /batch, /bulk or even / and we will have to add a uri and replace the id value by something provided by the consumer: Actions number 1 is DELETE /resources/ID1 and its result will be identified in the 207 response by the id ACTION1.
To see a complete example you should take a look at Facebook’s Graph API batch endpoint documentation. Note that this batch endpoint match request/response based on index and does far more than just processing a bunch of request.
https://www.codementor.io/blog/batch-endpoints-6olbjay1hd
Google has implemented a complicated but flexible batch endpoint. Instead of having an endpoint that accepts multiple resources, there's an endpoint that accepts multiple requests. These are essentially "meta" HTTP requests, where the main request contains different sub-requests.
Nate Barbettini – API Throwdown: RPC vs REST vs GraphQL, Iterate 2018
Best Practices for Building API Integrations
The difference between an API and an integration is that an API integration is how your application connects to an API.
WishTender Code Notes:
I've been confused about app architecture when your app gets complex.
It's easy when the routes map simply to a resource.
For example, adding a product to a shop.
In this case you might have a
Product.Model.js- a mongoose model. This is your resource.ProductService.js- a service that interacts with that model.product.js- a router that directs the express app based on different routes likeGET /product/213. The routes tells the app which service to use and what the response should be.
Thats simple.
But what about something not so straight forward?
For example, what route, service, and resource do we use when a client checks out? There isn't one clear resource here. When a client checks out we are interacting with many resources and services.
These might be:
- The
Order.Model.js, a mongoose model that stores order information such as cart items, time purchased, and currency conversions. - The shop owner's account
Owner.Model.jsbecause we might want to track their earnings on their account. - 3rd party API's, like Stripe for payment processing.
- An email service-
Email.Service.js- to send a reciept to the buyer and a notification to the shop owner.
and on and on...
In this case should we make a checkout.js route that maps POST /checkout to a CheckoutService.js?
CheckoutService.js could then interacts with many services and resources like
StripeService.jsOrderService.jsOwnerService.jsEmailService.js
Is this how it should be done?
A little more simply, what about when we're just creating a resources on a third party? For example a Stripe Connect account. Stripe Connect accounts: when you use Stripe as a marketplace, your platform has an account with Stripe and your users- like shop owners- have their own accounts called Connect accounts, connected to your platform account.
Your users need to create these Connect account. What route do you use? What service do you use?
It's not as simple as blogRoute.js--> BlogService.js -->Blog.Model.js. You don't house the resource on your site, the connect account object is housed on Stripe.
You might also create a model, StripeAccountInfo.Model.js to house some information about the account that is pertinent to your app- like currency, if the user paid their monthly fees to cover the Stripe Connect fees, and Stripe account number. But now, you have a resource on your site and a resource on stripe that need to be interacted with. How will this look like in your app architecture?
These are some things I've been confused by. I rarely see any information on these more complex circumstances. If any one has any resources, answers, is willing to talk it out over the phone, or even clues to what I should google I'd love to hear from you.
Twitter: @DashBarkHuss
LinkedIn: in/dashbh/
I reviewed circuit basics Lesson 1 - Voltage, Current, Resistance (Engineering Circuit Analysis) - really nice explanation of circuit basics.
WishTender Code Notes:
Working with the data in your app can be confusing when the data needs to be transformed into different representations.
Like prices.
- When you display prices they will look like $5,000.00. This will change depending on the currency.
- When the user inputs a price it might look like 5000, 5,000, 5000.00, 5.000,00 .
- And when you do calculations with prices you usually work in integers so $5,000.00 would be 5000000 pennies.
Keeping track of all these different formats can be confusing. How should we ultimately store prices in the database? When do these transformations happen from database format to view format and back?
I asked on StackOverflow:
Where should data be transformed for the database?
The answers showed there isn't one standard way. While sanitization happens on the backend for security reasons, normalization can happen in either place. People gave good ideas for how to decide. Check out the answers by @nickang, @helgedrews, and Cosmin Ioniță on stackoverflow here.
My question asked about transformations from the frontend user input to the backend. But I also wonder about transformations happening when data is sent from the backend to the frontend.
This answer suggests that the conversions should happen at the boundaries of the app. Unit conversion of data:
Keep consistent units within your system, and only convert to the user's preferred format at the boundaries of your system. The “boundary” doesn't necessarily mean front-end or back-end, this is more about how you structure the logic in your application. A classic MVC application would convert incoming data in the controllers, and outgoing data in the views.
I'm not sure if "incoming data" here means to the server or to the client. I'm going to assume incoming means from the frontend to the backend.
The same answerer above suggested:
Keep consistent units within your system...
Do not store the data in multiple formats in the database. This makes it so much harder to keep the various versions of the data consistent, and will make queries more complicated. Internally, your system should define a single authoritative data model.
I found a lot of ways people store price in the database, but the most common was to store price as an integer- the smallest unit of the currency. Here @helgedrews points to other API's for inspiration:
Furthermore, it can be helpful to take a look at some different APIs of well-known providers and how these handle prices.
The Paypal API uses an amount object to transfer prices as decimals together with a currency code.
"amount": { "value": "9.87", "currency": "USD" }It's up to you how to handle it in the frontend. Here is a link with an example request from the docs:
https://developer.paypal.com/docs/api/payments.payouts-batch/v1#payouts_post Stripe uses a slightly different model.
{ "unit_amount": 1600, "currency": "usd" }It has integer values in the base unit of the currency as the amount and a currency code to describe prices. Here are two examples to make it more clear:
In both cases, the normalization and sanitization has to be done before making requests. The response will also need formatting before showing it to the user. Of course, most of these requests are done by backend code. But if you look at the prebuilt checkout pages from Stripe or Paypal, these are also using normalized and sanitized values for their frontend integrations: https://developer.paypal.com/docs/business/checkout/configure-payments/single-page-app
So an outgoing boundary could be in the callback of a HTTP request. Here we do the transformation there before the client displays the data:
useEffect(() => {
if (!cart) {
fetchGet("/cart", (crt) => {
parseCartPrices(crt); // parseCartPrices transforms cart prices. ex: 100000 -> "1000.00"
setCart(crt);
});
}
}, [cart]);You could have this transformation go even further, 100000 -> "$1,000.00", or even 100000 -> "CA$1,800.10 estimated from $1,000.00". But I thought leaving it at "1000.00" would be more flexible. There are some calculations I do with the prices on the frontend. If the price was in the format "$1,000.00" I would have to parse it into "1000.00" before doing the calculations. So instead, I will further transform "1000.00" to "$1,000.00" only where necessary.
An incoming boundary could be in the onSubmit function of a form.
const onSubmit = (data) => {
data.price = toSmallestUnit(data.price, clientCurrency); // toSmallestUnit turns "100.00" into "10000" if currency is USD
props.onSubmit(data); // props.onSubmit sends the data to the backend in a POST request.
};Alternatively, you could transform "100.00" into "10000" just on the other side of the boundary in the backend, for example using an express-validator customSanitizer.
WishTender Code Notes:
You can handle some validations in your user model using Mongoose. For best practices, we want to make sure validation happens before business logic.
- Shailesh Shekhawat in How to make input validation simple and clean in your Express.js app
What is the best way to store monetary value in a relational database system like MySQL?
Storing prices in SQLite, what data-type to use?
Should I be serving price data up as integers, the way it's stored in my database (ex instead of $100.00, serve it as 10000)? The the front end has to transfrom it from 10000 to $100.00. Or should I transform it to $100.00 on the back end and serve that?
It gets more complicated when we take price conversion into account. If the client is in EUR and the price is in USD we have to do an estimate in the UI (ex "€120 estimated from $100"). That means we have to fetch the exchange rate and to the conversion on the frontend.
Should data transformation be on the front or on the back end in this scenario?
Very helpful: When is it appropriate to do calculations in front-end?. You could see many people diagreed but people gave reasons as to why you would do calculations on the front, back, or both.
Keep consistent units within your system, and only convert to the user's preferred format at the boundaries of your system. The “boundary” doesn't necessarily mean front-end or back-end, this is more about how you structure the logic in your application. A classic MVC application would convert incoming data in the controllers, and outgoing data in the views.
Do not store the data in multiple formats in the database. This makes it so much harder to keep the various versions of the data consistent, and will make queries more complicated. Internally, your system should define a single authoritative data model.
WishTender Code Notes:
I finished my sample currency i18n project. It's a sample project that shows how to set the currency for a user depending on their language preference in their browser accept-language header.
I wanted to accommodate different currencies because some of my beta users asked me to. But there was little info on the subject. I figured it out and put this sample together so others can understand how it might be done.
My app was re-rendering too many times. React doesn't have a solution to figuring out what caused a component to update (if you know of a way let me know). There are some notes in my last log entry (12-29-20) for some clues on how to do this- profiler, memo, react dev tools. I didn't find that any were very good for seeing why a component rendered. For example, memo only shows what props caused a re-render but not state changes.
So I built my own function to figure out when props or state changed. It doesn't show you when a parent rerender caused a component to rerender but you can figure it out if you use the debugging function in the parent and see that the parent re-renders.
This tells you if a component rendered and if the state or props changed.
It helped me figure out why I had so many re-renders. This function could be more DRY.
import { useEffect, useRef } from "react";
import _ from "lodash";
/**
* tracks component renders
* @param {String} component component name ex: App.name
* @param {Object} props
* @param {Object} state
*/
export default function useTraceUpdate(component, props, state) {
const prevP = useRef(props);
const prevS = useRef(state);
let render = useRef(0);
useEffect(() => {
render.current++;
console.log(component, "rendered ", render.current, " times");
if (props) {
const changedProps = Object.entries(props).reduce((ps, [k, v]) => {
if (prevP.current[k] !== v) {
ps[k] = [prevP.current[k], v];
}
return ps;
}, {});
if (Object.keys(changedProps).length > 0) {
const displayText = Object.entries(changedProps).map((s) =>
typeof s[1][0] === "object" && typeof s[1][1] === "object"
? `${s[0]} object changed, content of object ${
_.isEqual(s[1][0], s[1][1]) ? "didn't" : "did"
} change ${
!_.isEqual(s[1][0], s[1][1])
? `from ${s[1][0]} to ${s[1][1]}`
: ""
}`
: `${s[0]} changed from ${s[1][0]} \nto ${s[1][1]}`
);
console.log(
"r:" + render.current + " info: " + component + " changed props:",
displayText.join("\n")
);
}
}
prevP.current = props;
if (state) {
const changedState = Object.entries(state).reduce((ps, [k, v]) => {
if (prevS.current[k] !== v) {
ps[k] = [prevS.current[k], v];
}
return ps;
}, {});
if (Object.keys(changedState).length > 0) {
const displayText = Object.entries(changedState).map((s) =>
typeof s[1][0] === "object" && typeof s[1][1] === "object"
? `${s[0]} object changed, content of object ${
_.isEqual(s[1][0], s[1][1]) ? "didn't" : "did"
} change ${
!_.isEqual(s[1][0], s[1][1])
? `from ${s[1][0]} to ${s[1][1]}`
: ""
}`
: `${s[0]} changed from ${s[1][0]} \nto ${s[1][1]}`
);
console.log(
"r:" + render.current + " " + component + " changed state:",
displayText.join("\n")
);
}
}
if (render.current === 1) {
console.log("r:" + render.current + ": initial render");
}
prevS.current = state;
});
}How to use useTraceUpdate:
function WishlistPage(props) {
const [alias, setAlias] = useState(null);
const [wishlist, setWishlist] = useState(null);
const [refreshWishlist, setRefreshWishlist] = useState(null);
const currentUser = useContext(UserContext);
const states = {
alias,
wishlist,
refreshWishlist,
currentUser,
};
useTraceUpdate(WishlistPage.name, props, states);
return <div>... blablabla</div>;
}
export default WishlistPage;