Really interested in the idea of dynamic hotpatching for software security.
This is fairly widespread at the OS layer, but I think there's opp for it to happen ubiquitously from the chip up.
The way that software has historically been built and distributed it is effectively impossible to keep things up to date.
eMudhra key gets leaked and now suddenly all your Electron apps have to update. A vuln happen in Go stdlib and suddenly thousands of independent binaries need updates. This is an impossible situation.
You need to be running something on the OS at the root layer that can hotpatch _all_ of the software above it. This implies a hierarchical system where a root layer hotpatches e.g., the JVM, which then hotpatches Rhino, which hotpatches a JS lib.
Centrally managed, instantly distributable, instantly revertable.
Greasemonkey for everything.
Who is working on this and can I invest?
This is fairly widespread at the OS layer, but I think there's opp for it to happen ubiquitously from the chip up.
The way that software has historically been built and distributed it is effectively impossible to keep things up to date.
eMudhra key gets leaked and now suddenly all your Electron apps have to update. A vuln happen in Go stdlib and suddenly thousands of independent binaries need updates. This is an impossible situation.
You need to be running something on the OS at the root layer that can hotpatch _all_ of the software above it. This implies a hierarchical system where a root layer hotpatches e.g., the JVM, which then hotpatches Rhino, which hotpatches a JS lib.
Centrally managed, instantly distributable, instantly revertable.
Greasemonkey for everything.
Who is working on this and can I invest?
9
0
5
975
@aboodman this has been possible for years but its more trouble than its worth generally speaking
1) the code paths arent the same
2) best practices > hopeful magic fix-alls
3) changing byte code on the fly fucks with all forms of compliance
1) the code paths arent the same
2) best practices > hopeful magic fix-alls
3) changing byte code on the fly fucks with all forms of compliance
1
0
1
384
@aboodman i doubt it
you're better off shipping middleware for most of these things, which is what people already do
you're better off shipping middleware for most of these things, which is what people already do
Oct 24, 2026 路 11:20 PM UTC
1
0
0
32