Rationale: See diffstat
When this module is imported, it ends up calling stty via term_width()
in util.human.io.table(). When this happens outside of a terminal, the
following message is sent to stdout:
stty: 'standard input': Inappropriate ioctl for device
Not importing this module avoids that.
Furthermore three is value in this module having minimal dependencies as
they might not be available when it does the checks.
Ref a1fed82c44b9
Previous version of this patch used 'patternProperties' but that would
only work with simpler ECMA-262 regular expressions are also valid Lua
patterns.
Fixes that the more fixed width columns there are, the narrower the
resulting table becomes. A right-aligned variable-width column at the
last position should always be flush to the right side of the terminal.
I forget why I wanted this, but it may allow doing things like pull
settings from the column, especially when the mapper function is reused
among many columns.
As an alternative to doing it in the mapper function. Could be useful in
cases where one may want to put the ellipsis in the middle or beginning
instead of the start.
Makes it easier to make out where the set starts and ends in cases where
it may get embedded and tostring()-ed in a log message.
{ } taken over from util.array for consistency with some other systems
syntax for Sets, e.g. Python
Arrays in Lua do use { } but since __tostring is often user-facing it
seems sensible to use [ ] instead for consistency with many other
systems; as well as to allow the {a,b,c} formatting to be used by
util.set without being confused with util.array.
Instead of a percentage, this allows you to specify e.g. `width="[N]p"`, where
a width="2p" will be twice the width of a width="1p" column.
Compatibility with the old %-based widths is preserved, and percentages adding
up to more than 100 are handled more gracefully.
To avoid a race where server-started fires before the promise function body is
run (on next tick), I moved server-started to fire on the next tick, which
seems sensible anyway.
Errors are logged, I'm not sure if we ought to be doing something more here.
I'm sure we'll find out.
Should be able to roundtrip trough serialization. Also nice for debug
and logging purposes where you might want more details than what the
__tostring method provides.
Enables util.serialization to turn Sets into a representation that can be
deserialized with an environment trick, i.e. `set{"a","b"}`. Also useful
for debug purposes.
This was implied with the "debug" preset and does nice things like turn
util.set objects into "set{a,b,c}" instead of the quite verbose thing
you get otherwise.
Improves serialization of function references especially, the built-in
default handling of that in util.serialization is not the most
informative. Now, along with the function metatable from util.startup,
the actual function can be found by filename and line number.
RFC 6120 states that
> If the initiating entity does not wish to act on behalf of another
> entity, it MUST NOT provide an authorization identity.
Thus it seems weird to require it here. We can instead expect an
username from the token data passed back from the profile.
This follows the practice of util.sasl.external where the profile
callback returns the selected username, making the authentication module
responsible for extracting the username from the token.
This allows token-aware things to access extra information about the
authentication, such as when the token is due to expire and the attached
custom 'data'.
