💾 Archived View for gemi.dev › gemini-mailing-list › 000222.gmi captured on 2024-03-21 at 17:08:58. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-12-28)
-=-=-=-=-=-=-
Ok, so I'm writing this for multiple reasons that center around one reason. I'm just going to get into it. Just be aware that I'm writing this not to be mean, but to make sure everybody here pays attention to every consequence each decision has. :) We can *not* let Gemini turn into the Web, period. We have to remember how the Web turned into a disaster and actively prevent that. If we are moving toward making full Virtual Machine applications that can do threading and sockets and UI and GPU rendering and all of this... crap... we have then turned into the web and are requiring browser implementers to basically create full operating systems, which will prevent small teams from ever being able to develop a browser for Gemini, which means only bigger companies would be able to do this, which will reduce competition, and then we have a duopoly on browsers *and the spec*, like with what the web is today. Imo, we have to remember that not *everything* needs to be in *one unified protocol*. This is the exact line of thinking that led to HTTP and the other Web technologies. Not everyone who is browsing is a content creator. Content creators aren't necessarily browsing when creating content. Why do these two things need to be in the same protocol? Similarly, we already have Operating Systems that can run native programs millions of times faster than any Virtual Machine. We do not need a virtual machine. (I haven't seen anybody moving this far yet, but I still want to make this point because it's important). If we want something like being able to edit content on servers, what's wrong with (S)FTP? or SSH? or SCP? If something's wrong with these, then fix it! We can make a new, better, and simpler FTP! Or we can make a more complicated FTP! Or we can make a more Gopher/Gemini-focused FTP! (imo, FTP is very old and complicated, that we could actually use a new protocol for this, and this is something I've thought about way before Gemini was even a thing.) We can use Gemini's certificate concept for this new protocol that would be like a Semi-Anonymous FTP user. This being a separate protocol will allow for a lot more than if we try to shoehorn it into Gemini. Why should this be in a separate protocol? Gemini is not suited for Binary. And binary might be important for something like this. Gemini's display structure is suited for content, not editing filesystems or files. There are probably other reasons, but these are the most important. I wanted to say to Solderpunk... it's important that you take input, but just make sure you put your foot down when it needs to be done. Because, if anybody's read the Mythical-Man Month, or seen how committees (especially web committees) design things, large groups of people trying to decide things is not the best thing. It can have detrimental effects on the protocol. I appreciate everything you are doing Solderpunk, because it's important. And to everyone else, I would say, don't be afraid to design your own protocols. Competition is a good thing, and your ideas/goals are very likely to be different from anybody else's. I also want to announce that with recent developments in Gemini, I'm
---