@cyberkm@ChristerEricson If you're saying ui "text and filenames" 👍
Else, if "ui text" and "filenames", for runtime, I would argue that even filenames should be replaced with IDs in your archive 😁
@ChristerEricson Limit as much as possible external dependencies, and don't let them spread through your API. And think about your API sooner than later. Clean, simple, dependance-free API are a breeze to use.
@ChristerEricson When applicable, separation between runtime and edition data structure/serialization/code: this ensure performant, compact, and tight runtime, while keeping good user experience (content people AND progs) with adapted tools/edition code/data structure
@incomputable@lefticus I was mostly trying to enhance the dramatic effect by being categorical, but after all, maybe not. I've been fooled in the past by being sure that this O(n!) algo was the problem, only to discover by profiling that a container was passed by value.
@wadetb Well, again, I'm no expert, but to my understanding, it is just a proxy with a cache... We're using ngnix servers, and it's pretty light. Priming this cache depends mostly on the amount of data, and the quality of the link, not the server software.