Yup, things can be a mess when you can no longer source a part with the same spec. I haven't read the proposed legislation yet but does it account for being unable to do repairs because e.g. no one makes 555 chips anymore? (Albeit those are relatively simple.)
You can delay the porting if you buy a bunch of spare parts in advance but that has inventory price overhead.
How much of that would be easier if detailed specs were available? Could someone throw a pin-for-pin compatible implementation of that hypothetical 555 (to continue your example) out of a small FPGA?
Also, the most annoying to replace and most complicated parts are usually the ones that run software, which creates another problem. I bet half of the people at my local hackerspace could build a replacement board for their washing machine or fridge from just doing a visual inspection of the original, but it doesn't solve anything if you don't know what code the original was running. Having to hack the original into giving you its firmware blob raises the difficulty of replacement significantly, and you'll probably be breaking some laws too if you try it.
I wish we would be able to force companies into revealing both schematics and all code running on a device when they sunset it. But I know it won't happen, because the companies will immediately say that some of that old software is still being used on new devices, and therefore it is an Important Trade Secret.
For a washing machine your local hackathon could probably get most of the functionality in software as well. For years they were simple mechanical timers. Fill until the full switch trips, then run the motor for so long. then pump the water out, then run the motor at a higher speed....
Of course computers and modern sensors allow for "dirt sensors" which if they exist will be much harder to interface to.
Detailed specs make everything easier. :) Even if you just have the part name but not necessarily all the schematic info for how it's used that will help a lot because you can usually then go find the datasheet for it and from there figure out how to replace it if you need to. Replacement might be hard or easy, it really depends on the component. Functional specs would help but aren't necessarily enough, because there's extra information in the datasheet like value and thermal tolerances that the circuit might be relying on, and analog vs digital matters too. For 555 chips, those are used in analog circuits typically as clocks, so the functional spec would be something like what clock frequency is expected on the output line and maybe the shape of the wave. A clock is trivial with an FPGA or microcontroller but you can't simply stick one in the same pins as those are for digital circuits, at least not without some converter help. You might be able to replace a big chunk of the analog circuit with a digital timer though.
Binary blobs definitely are annoying. If the circuit is using a microcontroller but you don't have the code it was programmed with then as you say it's not exactly helpful that you could buy a blank replacement. But depending on the part you can sometimes get away with not knowing the code, or even being able to dump the code, and relying on capturing raw signals with an oscilloscope (assuming you have access to a functioning device) to develop a replacement. A lot of repairs don't always leave the thing just as good as before, so it might be acceptable to have a washing machine that only does a basic cycle because no one has yet deciphered the signal sequence to make it do other types. Or sacrifice the chime functionality. On the other hand modding to make a superior replacement is sometimes possible...
The only legislation I really like is not being able to go after modders/hackers/reverse engineers/repairers, even those making a profit as a side business. Mandating some sort of standard to make those things not exceed a certain level of difficulty seems really hairy to define and undesirable in general if it means a lot of things simply won't get made at all. Mandating opening after some time period or when a device is sunset might be workable and not impose any real expense but I don't see it being implemented any time soon. For binary blobs even if it's old or not a "trade secret", it's still "IP". (FPGA companies even call their programs "IP Cores".) We can't even get Green Eggs and Ham into the public domain for like another 40 years thanks to the 1978 law changes, so I wouldn't count on government force to get companies to open software up. The best approach seems to be the combination of cultural practice within software and slowly more and more of hardware communities, and pointing out the business incentives for sharing more things.
I didn't say it actually stopped, it was just the first chip that sprang to mind as an example chip, for good reason. Guess I should have seen this coming.
You can delay the porting if you buy a bunch of spare parts in advance but that has inventory price overhead.