Deployment got harder as their platform shifted beneath us, and one day recently our entire site was replaced by a simple message that said, "We no longer support Go 1.9." That was a fun day!
@snap_as was also the first product we launched separately from @write_as, and it was built on the assumption that an entirely separate codebase would be best.
The idea was that the applications would share common components (https://github.com/writeas/web-core), but would have separate application codebases and storage.
This might've been a good theory for building modular applications (as I'm hoping to), but in practice causes a ton of headaches for maintenance and end user experience.
Little problems came up right away:
- How do we share user data across platforms?
- How do we create common UI patterns while keeping distinct branding on each platform?
- How do we reconcile slightly-different uses of those shared libraries across platforms?
So, starting with @submit_as, we built the functionality into the Write.as codebase. There was still plenty of copy-paste involved, but we were able to launch this new product much more quickly.
Now instead of creating separate apps first, we're building them into Write.as first and then separating out functionality over time.
@writeas_dev @snap_as @write_as @matt First time I’ve heard of your app. Will check it out!
@plumey Awesome -- would love to hear feedback you have along the way, if any!
@writeas_dev @snap_as @write_as does this mean it will make its way into writefreely?
@liaizon Yep! Overall, now I'm thinking that all the "separate apps" I wanted to build make the most in a single codebase (#WriteFreely). Then admins can choose which "apps" to enable.
Either way, we're aiming to support photo uploads in WF by v1.0
@writeas_dev @liaizon Awesome 👍 I guess time for me to come back to extending @webimg to have some image housekeeping functions (in addition to watermarking and resize) and maybe @writeas_dev would find it useful for image uploads functionality in #writefreely 😊
A little background: this is driven by Google Cloud being a very, very bad cloud platform to build on as a small team.
We tried it in 2016 as an experiment, and a series of service deprecations since then have caused us downtime and meant us ultimately offering a stable, but stagnated service.