I've recently started to dabble in webdev (front/backend) and it's very much making me genuinely want to jump off and do _any_ other work because of the "endless cycle of webdev" and this is despite loving the concept of it (it has made me me understand previous hypes such as having open APIs like in early 2010s since I am in that phase/realization).
There are 2 problems I have personally with it and 1 is a subjective observation:
1: It harms the ability to learn the concepts of webdev (routing, API routing, hydration, SSG/SSR, templating, etc.) - due to:
- The frequency of concepts being reinvented or considered obsolete (Example: REST API -> GraphQL -> trpc)
- There is no well established "timeless book"/webdoc that explains these concepts (haven't found it at least)
2: it's demoralizing:
- Making you second guess the stack/libs you've chosen and forcing you to relearn and thus never making that "prefect" stable code
- Being told years later that the entire concept/tool has been a disaster for the web
Subjective:
3: Established "old" frameworks are considered harmful/bug-riddled/slow with the only option being to move over to the new trendy framework thus causing the cycle to occur once more.
----
2, It's partially why I refuse to use/learn Typescript (the main reason is that Microsoft controls it - in contrast with say Go or Rust where it's controlled by a foundation), due to the fate of coffescript and other web trends that now has tons of "x was a mistake" (like SPA).
3, this is mostly a rebuttable to the response "just keep using what you like",
because I've been burnt before using what was seen once as well-established and polished but when I tried it, it was in total decay (meteorJS).
Essentially I'm conflicted on what to do, I do like the ideas of webapps - that webapps can connect other services together or able to create 1 platform for "every" device.
But the ecosystem is all over the place and where it's very much "chase the unicorn" make huge claims with almost zero long-term plans.
Lastly, I'm not against moving to new tech or new stuff, I very much welcome it, if it is swapping out 1 small library (lets say something the size of sentry.io or prisma.io) not the entire workbench.
There are 2 problems I have personally with it and 1 is a subjective observation:
1: It harms the ability to learn the concepts of webdev (routing, API routing, hydration, SSG/SSR, templating, etc.) - due to:
- The frequency of concepts being reinvented or considered obsolete (Example: REST API -> GraphQL -> trpc)
- There is no well established "timeless book"/webdoc that explains these concepts (haven't found it at least)
2: it's demoralizing:
- Making you second guess the stack/libs you've chosen and forcing you to relearn and thus never making that "prefect" stable code
- Being told years later that the entire concept/tool has been a disaster for the web
Subjective:
3: Established "old" frameworks are considered harmful/bug-riddled/slow with the only option being to move over to the new trendy framework thus causing the cycle to occur once more.
----
2, It's partially why I refuse to use/learn Typescript (the main reason is that Microsoft controls it - in contrast with say Go or Rust where it's controlled by a foundation), due to the fate of coffescript and other web trends that now has tons of "x was a mistake" (like SPA).
3, this is mostly a rebuttable to the response "just keep using what you like", because I've been burnt before using what was seen once as well-established and polished but when I tried it, it was in total decay (meteorJS).
Essentially I'm conflicted on what to do, I do like the ideas of webapps - that webapps can connect other services together or able to create 1 platform for "every" device.
But the ecosystem is all over the place and where it's very much "chase the unicorn" make huge claims with almost zero long-term plans.
Lastly, I'm not against moving to new tech or new stuff, I very much welcome it, if it is swapping out 1 small library (lets say something the size of sentry.io or prisma.io) not the entire workbench.