This change avoids the problem of a bad "index.html" on a doucment
rendering error, where the presense of an "index.html" would count
as successes on a retry.
The exit handler in the user process flushes via the plumber,
but maybe DrRacket could do more with the user's plumber, such as
attempting to exit instead of always forcing the old user process to
terinate when the "Run" button is clicked.
A sandbox needs it own plumber so that sandboxed code cannot attach an
arbitrary callback to the plumber outside the sandbox. The plumber
outside the sandbox gets a flush callback that tells (within the
sandbox thread) the new plumber to flush.
In v6.0.1.7, I tried to give a port-flushing job to custodians. They
turned out to be unqualified, so let's try employing specialists.
Thanks to Eli for pointing out the problem with the v6.0.1.7 design:
attaching callbacks to custodians allows a sandboxed task to escape
through the custodian hierarchy. Plumbers avoid this problem by
having no hierarchy.
The `catalog-archive` command not only copies a catalog, but also copies
all package sources referenced by the catalog to create a snapshot that
doesn't depend on any external pieces.
One reason you might want to do that is to download all packages on a
machine with general network access, and then building packages on a
(virtual) machine whose network access is limited.
This commit also adds support for relative package sources in the
catalog protocol, where a client converts a relative package source to
an absolute one. Since the relative-to-absolute conversion happens on
clients, relative sources will be of limited use until v6.0 and v6.0.1
become deprecated.
A new `current-pkg-lookup-version` parameter corrects the misuse of
`current-pkg-scope-version` for package-catalog queries.
This change solves a long-standing problem that reencoded output
was not flushed on exit, especially in the case that the
current output port is reencoded.
This provides approximately 6% speedup on
`racket -l math/scribblings/math.scrbl` and about
14% speedup on the `new-metrics` test.
Mostly this involves threading #f through the whole
of the inference process. There are several new
macros in `typed-racket/infer/fail` which are useful
for comprehensively using Maybe-monad style
programming in Racket. Of particular note is `%`,
which satisfies (% f e ...) => (and e ... (f e ...))
but with the obvious fixes.
This commit also weakens several contracts which
caused the build of DrRacket and/or `math` to fail
when contracts were enabled.
