I took time off to watch the new Podlodka season with full strength (actually not, it just coincided). The automation week really hits me right now, so I have a lot of thoughts on each talk, I'll try to record something each day for myself.
The first talk of the day from Nikita Kulikov was about Github Actions. And it's actually quite sad. Because I already have anxiety about how much my life is tied to GitHub, and Actions is too good not to choose it in almost any situation over other popular CIs. And it's cheaper, sometimes even free, and there's a marketplace of actions that are always missing in all sorts of GitLabs. It becomes too easy to write the yamls themselves, due to which you sink deeper and deeper into this swamp of dependency. It always seemed to me that the rule of good manners for CI is to transfer as much complexity of the setup as possible from the CI solution to the project side, in our case write Gradle tasks and all that. But if there's almost no setup complexity at all, then maybe it's not so scary. Well, and I remained of the opinion that writing CI configs is not difficult, it's difficult to build such a behavior model that will work effectively on a specific project.
The second talk from Nikita Yatskivsky was about automating token export from Figma. I'm literally doing this right now and it was interesting to compare experiences, although it didn't go far from the original article on Habr. I don't have the feeling that all such solutions are made from crap and sticks without the possibility to normally reuse between projects. Automation consists of two parts: export from Figma API to some intermediate format and generate code from this format. And the problem is that in Figma the node structure is different for everyone and everyone needs different code, hence the not invented here syndrome. So it's interesting to listen about theory, but in practice you still have to do everything yourself, fortunately this is a maximally straightforward activity. It's a pity that the most interesting thing for me remained hidden behind the words "Figma didn't give us enterprise because we're a Russian company," and variables are exactly what can potentially protect us from different Figma structures, facilitate reusability, and save us from designers' mischievous hands.
I hope there will be a little more insights and guts further. So far everything is quite high-level and helps perhaps only to validate your adequacy or trigger reflections on all these topics. But if you have experience, you don't learn much new, although the guys objectively cooked very well in the topic. More interesting stuff in the chat, as always)