It is dangerous to set this on a per-process basis. If we even
still want the ability to have observer errors not kill the
session (debatable) we should do so by having an option on the
session object that is set when the shiny server function runs.
Not only does this remove support for the option, but now we
will refuse to connect and will abort the session if the option
is set to a non-NULL value. This is to prevent apps from moving
forward with the assumption that their option will work, when
it really won't. (I doubt anyone is using it anyway, it's so
obscure...)
- More selective removal of frames at end of call stack
- Add withLogErrors, printError, stripStackTrace convenience funcs
- Properly capture/log errors for various levels of unhandled errors
- Unhide stacks for flush/flushed/sessionended callbacks
These functions were created before getDefaultReactiveDomain()
existed, so the only way to get ahold of the current session was
if the caller explicitly passed it.
This is slightly backwards incompatible, in that existing calls
to invalidateLater() that don't pass a session argument will
behave slightly differently (bound to the current session instead
of to no session), but those calls would have triggered a warning
for all but the very earliest versions of Shiny.
Instead of showing the code, try to just show the name
of the reactive/output. Uses a fairly flaky algorithm
for determining the name of the reactive; will only
work in cases where the definition of the reactive
begins with "foo <- reactive({".
- Need to carefully review ShinySession, think about
each and every public method and whether it should
be passthrough or do something different for a
namespaced session.
- It would be nice if we do the namespace splitting
at the server.R level and actually have numerous
reactivevalues objects for inputs, one for each
namespace. This will make the reactive hooks for
names() and reactiveValuesToList() correct, as
right now they will oversubscribe if you're in a
namespaced session. Also it would prevent outer
sessions from seeing the values of their child
namespaces. (Though is that good or bad...?)
After discussing with @trestletech, the semantics are just not right.
When using it with a renderXXX function, it doesn't actually stop
recomputation, so the only reasonable place to use it is in a reactive
expression, but if that's true then this is a terrible looking API.
Will replace soon with something better.
- Unhandled observer error no longer stops flushReact in mid-flush
- Observer errors are now warnings that include the observer label
- Session is explicitly being closed by Shiny instead of by httpuv
Also:
- session$close() is now public and closes the session
- Introduce randomInt/p_randomInt to generate random integers in a half-open range
- Stop using runif to generate integers
- Explicitly reset the private seed during .onLoad. I was getting the same "random" numbers from Shiny every time I restarted R!
