I detest this piece of crap idea so much. You've created "files" and made an interface to them, but they are in no way usable or empowering to users? The files don't really exist at all? They are only for devs to use? How have you fucked up files so badly? Why would you do this? This is just an absolutely ruthlessly awful idea, deeply anti-user, a perfect shit-storm of user-loathing.
On the plus side, at least the spec authors didn't build a ridiculous overprotective childs toy like the File System Access api. Which does actually let the web work with files users can also work with. But which makes all kinds of huge promises about durability & robustness that make sure that the actual API webdevs get for dealing with files is ridiculously hideously unperformant & riddles with problems so as to be nearly useless. It's unusable for any kind of interesting application like running sqlite or any other scenario where data is updated in-place. Deno evaluated this API & basically laughed it out of the room as infeasible & worthless, on a number of dimensions. https://github.com/denoland/deno/issues/11018
So the web is in the worst possible spot. We have one API where we can do nothing for our users: we can build fancy new data tools with Storage Foundation but we can't work with user files or expose what we're doing. Or we can work with File System Access, and manipulate user files, but incredibly slowly. I think of Adobe recently trying to port Photoshop & Illustrator to the web, & how much they must be cursing the "hell" or "damnation" options they've been given with this sickening situation.
There is work to try to take much of the Storage Foundation API & to extend & build it out into Yet Another Specification, currently called Access Handles, that does interface with real files. This is the only chance we're going to have in the next half decade for the web to not be utterly trash at working with files. Pray for the suck to stop.
I had this thought too when seeing Adobe's announcement - I would have thought the lack of a good filesystem API would block them more than even porting GPU intensive workloads to the browser
On the plus side, at least the spec authors didn't build a ridiculous overprotective childs toy like the File System Access api. Which does actually let the web work with files users can also work with. But which makes all kinds of huge promises about durability & robustness that make sure that the actual API webdevs get for dealing with files is ridiculously hideously unperformant & riddles with problems so as to be nearly useless. It's unusable for any kind of interesting application like running sqlite or any other scenario where data is updated in-place. Deno evaluated this API & basically laughed it out of the room as infeasible & worthless, on a number of dimensions. https://github.com/denoland/deno/issues/11018
So the web is in the worst possible spot. We have one API where we can do nothing for our users: we can build fancy new data tools with Storage Foundation but we can't work with user files or expose what we're doing. Or we can work with File System Access, and manipulate user files, but incredibly slowly. I think of Adobe recently trying to port Photoshop & Illustrator to the web, & how much they must be cursing the "hell" or "damnation" options they've been given with this sickening situation.
There is work to try to take much of the Storage Foundation API & to extend & build it out into Yet Another Specification, currently called Access Handles, that does interface with real files. This is the only chance we're going to have in the next half decade for the web to not be utterly trash at working with files. Pray for the suck to stop.