You know how every Windows program begins with a small DOS program that says “this can’t be run in DOS mode”? What if you replace that with a small HTML page that loads a wasm VM for the executable’s real target environment
-
Show this thread
-
Replying to @jckarter
I believe this is precisely the plan, modulo some window dressing language about “graceful degradation”
2 replies 0 retweets 4 likes -
Replying to @graydon_pub
I can’t wait until twenty years from now when our firmware drivers are all required by spec to include a “This website requires Google Chrome” stub
1 reply 4 retweets 32 likes -
Replying to @jckarter
In fairness to the plan, the existing accreted web platform targeted for replacement is kinda an engineering catastrophe. I think replacing it with another mobile code thing is unwise; but it’d be hard to make a bigger mess than the present.
2 replies 0 retweets 5 likes -
Replying to @graydon_pub
The “mobile code” thing seems like a big part of the success of the platform, though. A lot of the web’s underlying substrate, like zero-install deployment, URLs as universally sharable state seeds, working back button, etc. haven’t been successfully replicated
1 reply 0 retweets 8 likes -
Replying to @jckarter
Personal opinion: the mobile code aspect has been very bad for it. I think a global client-server form interface and a global static hypermedia system are both totally sellable on their own merits w/o mobile code.
2 replies 1 retweet 8 likes -
Replying to @graydon_pub
History seems to suggest that every platform that isn’t designed to be Turing complete from the onset evolves to become Turing complete, badly or by accident
2 replies 0 retweets 10 likes -
Replying to @jckarter @graydon_pub
The discipline required to avoid being accidentally turing complete is simply impossible for any process that allows experimental enhancements. Only way I can see to maintain it is a test suite that requires a formal proof that you haven’t made it TC. But that’s elitist.
5 replies 1 retweet 4 likes -
There are really simple approaches to avoiding Turing completeness. For example forward-jumps-only.
2 replies 0 retweets 2 likes
You can expand this to be more practical by allowing forward calls too (with return back to call point), which also yields bounded loops (via unrolled sequence of calls), etc.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.