For instances where svc_fs.node was called and explicitly passed a
NodePathSelector (instead of being passed a legacy-style object
with a path attribute) this code was failing. This commit moves the
affected code snippet after input coercion and updates references
to use the coerced input.
style: formatting changes were applied to the code. Instead of using
nested conditional blocks we use labelled breaks
Not sure when this went away, but we were always getting the
permission though the file path instead of ID/UUID from what it
looks like, which was hiding this problem.
Add a simple 20 second cache with kv.js to the scan_ method of
PermissionService. This will not improve the performance for
the permission system, but it will invoke it less often.
This is a quick solution for javascript files that should not have Puter
extension code prepended to the top of them. We may want to do something
different in the future like skipping directories named "gui" or
following imports (we can only do that with ESM though) or actually
having a way to run ES modules in their own global context (which is
avaialble in node.js as vm.Module under an experimental flag, but there
is no way to use the default linker for imports in the module);
basically there's no good way to do this and it is very unfortunate.
Previous commit to do this contained accidental other changes that I
staged and forgot about. This commit really only contains the sync for
package-json. This was broken for a while - since when we did work on
AWS secrets manager.
* feat: Differentiate tablets by pointer type
Reclassifies tablets with a mouse/trackpad to use the desktop UI.
This is detected by checking for `(hover: hover)` media query capabilities. This ensures that touch-only tablets retain their original, touch-friendly mobile UI while fixing the bug for users with pointer devices.
* Make sure `matchMedia` exists before using it
---------
Co-authored-by: Nariman Jelveh <nj@puter.com>
This error message in UIDesktop caused me confusion earlier because it
didn't provide any indication of where the error came from. This commit
adds a string behind the error to specify that it happened while trying
to launch an app.
This commit adds an undefined check for metadata from Puter's desktop
GUI inside a filesystem operation frame; this information won't always
be available.
Allows disabling abuse checks undefined origin checks with config
parameters to make development more convenient. The approach before this
change has been to manually comment out these lines of code.
The extension pseudo-global was only available at the root synchronous
execution context of a module loaded as an extension, which meant it
wasn't available for such things as handler functions for listeners even
though it looks like it should be lexically scoped.
This commit writes a `const` to the top of javascript files in the
runtime copy of an extension so that it works as nearly to a real global
as possible.
A real global isn't possible here for two reasons. First, running in the
same v8 VM as Puter means we can't have distinct global objects with the
same name across different extensions. Second, we can't use node:vm
because we would then be unable to support ES modules, which are
experimental in node:vm and also there's no equivalent to createRequire
for the linker even if we wanted to require the
--experimental-vm-modules flag in our package.json. This seems to be the
best we can do given the state of affairs of the runtime.
Some monitoring services want a system to look like it's failing hard
instead of a system being able to report that it's failing successfully.
This commit adds an alternative to healthcheck that throws 500 when any
service is failing, as though reporting such depended on the services
working, to appease these monitoring services and their semantically
incorrect nature of operation.
This commit message sets the precedent for "mon:" as a conventional
commit style prefix for changes related to system health and monitoring.
This optimization speends up load time for Puter significantly when many
extensions are present. This is also necessary for subsequent changes
which may require prepending code to all the source files inside an
extension, which would be too slow without this optimization.
We can assume Arch users are comfortable with making configuration
changes and observing high volume log output (maybe not true for
alternative distributions of arch, but for base Arch absolutely) so it
probably makes sense to have the log minimizer off for these cases.
KernelDeimos