Interesting challenge trying to build a locale-specific application that the whole world seems to want to use. I have users across 27 countries; time to get busy.
New sign ups for 1 week straight on a not marketed website, very cool!
In other news im feeling scaling pains in my 150k loc codebase. CI bill is skyrocketing, AI code review bill is skyrocketing, going to have to make money pretty quick at this rate
getting alot of hits from chatgpt on a project ive been working on for around a year
its not ready yet, i havent marketed, i was actually thinking i was heading into a time where i could overhaul alot of the project to really make it shine how i want
but the users are here. I had to create a gate to block sign ups. last night i was testing onboarding flows so i opened sign ups for myself to test, fell asleep and forgot to close the gate
woke up to more sign ups, idk where im going with this, i guess we are locking in
Continuing to researching what makes a text buffer.
It started out as just a blanket statement. "create text buffer" and at this point I've answered the big unknowns; what is a text buffer? what does the spec look like? what conventions can i reference in my building? With this foundational knowledge it's turning into just another coding challenge
I discovered where i SHOULDN'T be sinking my time too. I started off thinking about data structures and algorithms and was trying to land on the "perfect" data structure but through that research I realized that the input / output are where the initial focus should be
I now have that "create text buffer" statement narrowed down to an abstract interface. I'm finding that documentation and tests are really easy to implement at this stage as I research what each operation should do.
like i'm just writing tests and docs but no code and probably wont get the opportunity to for a few days. Not the worst thing in the world for this because i'm getting time to think and sleep on decisions, which isn't always what i get to do at my day job.
Just a nice change of pace really, i'm enjoying the learning journey and am on pace to deliver my best quality work yet
btw i'm using a string as the initial data structure for the text buffer, will use that as a baseline benchmark for performance and circle back to more performant solutions after there is something to experience
Day 8 of working on this markdown editor. (Definitely not my full time focus.)
So far I know I need a text buffer. I started with a hand wavy AI generated definition of done and have been refining that definition as I continue to research and learn more about text buffers
No code yet. Ive been looking at abstractions in the wild like ropey for example. Still trying to get an idea for what i want the sdk interface to look like, AND comprehend what needs to happen
I cant hand the implementation off to AI here because its too far out of my realm of expertise. I need to have my own opinions before I can let AI drive
A cool example of these abstractions can be observed in https://t.co/Scfltqjnt6
that whole app is 2 abstractions and a database (seating canvas component, internal auth component lob)
so the win is that the seating canvas app lives in its own package, which gets imported across multiple environments (Storybook /Vite, Next.js) and used in multiple projects (Storybook, https://t.co/Scfltqjnt6, https://t.co/np5f9JNqwM (soon))
Building marriednext i went 10x engineer and created like 3-4 full featured products within marriednext.
I decided to step back abstract each feature into a standalone app. Its been one of those eras where it looks like no progress is being made from a product facing perspective
the effort is starting to show. ive soft launched a alpha version of https://t.co/ScfltqiPDy, working on alpha releases of 2 more adjacent products under different domains.
im also doubling back and closing in on the relaunch of https://t.co/np5f9JMSHe, where im integrating all 3 domains under the marriednext domain to cater to weddings specifically
Spending about 30 min - 1 hr a day on my markdown editor right now while i wrap up other projects
ive basically been spending time getting my feet wet in with specific data structures common in text buffers (PieceTable, PieceTree, Rope and more that turned out to be irrelevant)
this whole time i have been trying to decide the algorithm I want to use;
also realizing at the same time that the algorithm doesn't even matter at this stage because the input/output is the same regardless
going to just use an array of strings to start and circle back later ; which is fine this will give me more time to research and mull over these data structures
used to lol at this price but now here i am actually weighing the risk / reward
looking at my current mac i've ran it daily since 2021
i don't doubt i would squeeze every penny out of this machine but man this hurts the wallet
i will be building a markdown editor in public , currently in the planning phase; but will begin to share regular progress updates
(expect more library author style technical posts, aka probably boring to the majority)
all i can say for now is; 0 external dependencies, and i'm thinking about AI at the ground level