The contract now has two major differences:
- It raises an error when it would have to wrap.
- It uses chaperones to delay errors as long as possible
In general, using `Any` as a type when exporting to untyped
code will now just work, unless the untyped code tries to
communicate values back to the typed side, in which case an
immediate error will be raised.
Much of the implementation comes from the membrane design
from [Strickland et al, OOPSLA 2012].
raw-module-path inside of a phaseless-spec (see
the #%require docs for the description of these).
Also, Rackety
in conjunction with commit 9047427 (and an earlier
commit in those files/dirs), this commit:
closes PR 7815
closes PR 10455
closes PR 10788
(the way things currently stand, check syntax needs more information
from the fully expanded form, but at least now it has a better chance
to actually use that information, if it were there ...)
related to PR 7815
related to PR 10455
related to PR 10788
An attempt to detect a submodule could trigger the original module
name resolver when the would-be enclosing module would be handled
by the embedding-specific resolver. When a submodule is not found
but its would-be enclosing module is embedded, then assume that
the default resolver wouldn't find the submodule, eithe --- and
therefore avoid a potential "collection not found" error.
Also, add 'lsquo as allowed content.
Omitting the ` conversion in the first place was over-conservative.
There's a backward-compatibility issue with this addition (i.e., a
document might contain a backquote in a decoded context that is
meant to be rendered as a backquote), but the potential problems
seem minor.
The `slideshow/code-pict' library is the same as `slideshow/code', but
it works in non-GUI settings. Only the `slideshow/code' library connects
the code font size to `current-font-size', though.
The `code' macro, `define-code', etc., now support "code transformers",
which are syntax bindings that trigger otherwise-unescaped transformations
in the code to typeset (which can make the code easier to read and
friendlier to auto-indentation).
Other major changes:
- pg code now uses only binary format
- pg timestamptz now always UTC (tz = 0), added doc section
- added contracts to most pg "can't-convert" errors
Support for break clauses complicates expansion to `for/fold/derived';
a new `syntax/for-body' library provides a helper for macros that need
to split a `for'-style body into a prefix part and wrappable part.
Allows the use of `in-generator' to produce multiple values in a
position other than immediately within `for' (where the arity
can be inferred).
Closes PR 11662
The new parameter (and supporting environment variables and
command-line flags) can bytecode lookup to a tree other than
where a source file resides, so that sources and generated
compiled files can be kept separate. It also supports storing
bytecode files in a version-specific location (either with
the source or elsewhere).
The `make-log-receiver' function now includes a logger-name
filter. This filter is implemented as a low enough level that
it affects `log-level?' tests to check whether a log message
needs to be constructed at all.
The -W and -L flags and PLTSTDERR and PLTSYSLOG environment variables
support filters of the form "<level> <level>@<name> ...", where
<level>@<name> specializes filtering of events for a logger whose
name matches <name> to show <level> and higher.
The old `cast' didn't work right for a mismatch between
a pointer GCableness and the source or target types, and
it didn't work right for an GCable pointer with a non-zero
offset. While those pitfalls were documented, the first
of them definitely has been a source of bugs in code that
I wrote.
Also added `cpointer-gcable?'
Add `file-position*', which can return #f instead of raising
an exception when a port's position is unknown. Change
`make-input-port' and `make-output-port' to accept more
kinds of values as the initial position.
These changes make it possible to synchronize a port's
position with a `port-commit-peeked' action. It's ugly,
which I think reflect something broken about position
tracking in the port protocol (which seems difficult to fix
without breaking compaibility).
Providing a port instead of a reading or writing procedure
redirects the read/write to the specified port. This shortcut
is kind of a hack, but the run-time system can easily streamline
the redirection when it's exposed this way.
Using the new redirection feature reduces overhead in
`with-output-to-bytes' and `pretty-print'.
We can't disallow the creation of bad mutators without breaking
old code, but we can prevent the JIT from treating them like
good ones.
Closes PR 13062
Stream generic operations stopped working for lists
since the operations used only the generic dispatcher
instead of the real generic functions.
(Moral of this story: write more tests)
that it drops from the expansion (like define/public) by
adding them to the origin syntax property (and sometimes
to disappeared-use; see the add-decl-props function
for details on those that aren't in the origin property)
this means that check syntax will now pick them up
so they'll show up in the blue boxes in drracket
Thanks Matthew, for some helpful advice and
comments on an initial version of the commit.
Generalize splitting of `(let-values ([(x ...) (values e ...)]) ....)'
to `(let ([x e] ...) ....)' for any `e', since it's always equivalent.
Right?
(The old requirements on the `e's seem to be needed only for
`letrec-values' splitting and maybe mutable variables.)
Treat unsafe functional operations (which never raise an
exception) as omitable, which means that simple `let-values'
combinations can be split into `let' bindings, etc.
Basically, Racklog (and all versions of schelog) implement ! by
causing the failure continuation of the entire relation being
returned. They did not also cause the unification caused by the
relation to be un-done.
However, it is not easy to separate un-doing the local changes because
the unification just returns a failure continuation too. I had to call
that fail continuation but use state to communicate to its target that
the next clause should not be visited.
I don't know if this is correct. My test suite contains a lot of cut
tests that still pass. Erik's test passes too. But I'm not confident
that this really works.
Internally, there's a `prop:method-arity-error' property that is
used for keyword-accepting methods. The same thing could be
accomplished with `procedure->method', but the new property avoids
a wrapper. It might be nice to expose the property from `racket/base',
but that creates trouble for generating arity errors for keyword-
requiring procedures (i.e., when such a procedure is wrapped), so
keep it provate for now.
Closes PR 12982
Split out base-abbrev.rkt so that subtype is not dependent on abbrev.rkt.
Remove unused code in numeric-tower.rkt so that it is now a dependent of
abbrev.rkt, which allows the body of convenience.rkt to be merged back in.
Remove special casing for union.rkt and extraneous subtyping checks.
Remove union-maker.
conventions in 9.2.1 of the reference (altho the messages do
not yet do the extra level of indenting when a field is too
long, nor are there any field names ending in ...)
Also, fix the docs for the #:stronger argument to
make-contract, make-chaperone-contract, and make-flat-contract
Specifically, it seems like about 20% of the time (in drdr),
running the program
(let l()(l))
in DrRacket and then clicking the break button results in a state
where DrRacket's focus is not in the definitions window. I can't seem
to make this happen on my own machine and I'm not sure if this a
race-condition in the test suite or a real bug in DrRacket but it
seems minor enough (given all of the other focus-based testing that is
happening in this (and related) test suites) that I'm just going to
give up on this particular test.
A progress evt from a close input port must be initially ready,
and the primitive `peek-bytes-avail!' checks a progress evt
before checking whether the port is closed.
These changes resolve a race in `read-bytes-evt' and related evt
constructors.
This is a follow-up to commit ec6f3fd610. We're still
seeing crashes while rendering the "plot" documentation, and this
change seems to make things work on my machine.
Fix tchecking for a rest argument to a function that
is lifted by closure conversion so that one of its
arguments is a mutable local variable's location.
Also reject bytecode that would pass too many arguments
to a lifted function, since that would trigger an arity
error that might try to use a location as a value.
Merge to v5.3
When a `port-commit-peeked' succeeds, position information should
(appear to) be updated. This patch synchronizes commits and
position information for primitive ports, but synchronizing
them for user ports remains a problem.
Convert
(hash-ref <hash> <key-expr> (lambda () <literal>))
to
(hash-ref <hash> <key-expr> <literal>)
which is useful for making the `case' expansion fit
Typed Racket.
appear in saved wxme format files
also, improve the testing support for testing snip loading
(before this, the testing infrastructure could let one test
"leak" into another one in a way that could mask failures)
please include in release branch
Leave it working in splicing mode. I prefer doing that over always
splicing them, since that would make a less uniform interface, so I
rather keep all options open. There is no longer a `#:nothing' keyword,
which is the main point of this downgrade.
(See mailing list discussion on "no-argument" for the reason.)
and the change to the racket/gui load-handler
(unfortunately, there is still another problem that keeps
the test suite from passing)
please merge to the release branch
When a module is loaded from bytecode and then the value of
`use-compiled-file-paths' changes, an attempt to load a submodule
would fail, because source isn't used if the main module is
already declared, and the bytecode code is not used according to
`use-compiled-file-paths'. Make the bytecode path stick when it
is used once, so that submodule loads succeed, and make it work
even with `namespace-module-attach'.
The module-attach part of this protocol requires a change to the
API of a module name resolver: the notification mode gets two
arguments, instead of one, where the second argument is an
environment.
The lambda-lifting transformation needs to iterate to a fixpoint
where each lambda's added arguments and order are known. The
check for whether something changed was formerly just the number
of added arguments, but that's not good enough, because a binding
might get lifted away while another one acquires an extra argument.
The right test is to check the count and original bindings for the
added arguments.
Closes PR 12910
- Allow indexing into a VectorTop, with result `Any`.
- Don't use special typing rules for applications when the operator
has an annotation or instantiation.
Closes PR 12887.
Closes PR 12888.
The optional arguments for `call-as-current' for `gl-context<%>'
were not implemented, and the locking implementation didn't match
the documentation in other ways.
The libraries moved were:
- mzlib/control => racket/control
- mzlib/date => racket/date
- mzlib/deflate => file/gzip
- mzlib/inflate => file/gunzip
- mzlib/port => racket/port
- mzlib/process => racket/system
- mzlib/runtime-path => racket/runtime-path
- mzlib/shared => racket/shared
- mzlib/unit => racket/unit
- mzlib/unit-exptime => racket/unit-exptime
- mzlib/zip => file/zip
The old modules in mzlib are now pointers to the
new modules. These are all modules that were already
redirected in the documentation.
Added empty-sequence type (prints funny but works polymorphically; will submit bug report)
Loosened type of sequence-andmap (can't mimic andmap's predicate type)
Paths are left as paths, instead of trying to convert them to strings
or byte strings. Submodule path elements should be unquoted -- in the
same form as a `submod' form. All extra parts are submodule path elements,
never module paths or ".".
Each typed module now defines a submodule named `type-decl`.
This module performs the type environment initialization (along
with other environment updates) when invoked. Additionall,
every typed module, when invoked, performs a for-syntax addition
to a list specifying the submodules that need invocation.
This invocation is then performed by the `#%module-begin` from
Typed Racket.
The `type-decl` module always goes at the beginning of the
expanded module, so that it's available at syntax-time for all
the other submodules. This involved adding pre- and post-
syntaxes for the results of typechecking.
This allows significant runtime dependency reduction from the
main `typed/racket` and `typed/racket/base` languages (not yet
complete).
Fixed problems related to sorting, more than two references for
one citation, and "specific" additions like page numbers.
Also, removed a set of parentheses around disambiguated dates
in the bibliography, because I don't think they belong there.
The doc format was confused; for example, square brackets don't mean
optional in a syntactic form documentation, but instead mean square
brackets.
The JIT wants to detect calls to functions bound to module-level
variables that are always instantiated to a procedure of a particular
arity, in which case it can avoid some checks. The problem is that
bytecode and JITted code can be shared via the module cache across
namespaces that use different modules to implement a particular module
path or that enable the JIT differently. In particular, starting with
one that has a procedure binding and then using one (in a different
namespace) that has a non-procedure binding could lead to a crash.
Defend against this possibility by never treating imported variables
as constant in that sense. The JIT detects imported variables through
a new import map in the prefix.
This change may slow code. My guess is that it will have no
performance impact in practice, due to cross-module inlining,
although I can construct a microbenchmark that slows by 20%.
The optimizer can no longer reduce
(with-continuation-mark _id _v-expr _expr)
to just
_expr
when _v-expr and _expr are simple enough, because _id
might be bound to a continuation mark key with an
impersonator that checks the result of _v-expr.
The loss of an optimization can have a significant affect on
errortrace of microbenchmarks, such as
(for ([i (in-range 10000000)])
i)
The new names are `prompt-tag/c` and `continuation-mark-key/c`
to keep the names consistent with the values that are being
contracted. Also updated the HISTORY file.
Add a `#:nothing' argument so the no-value value can be
made explicit --- based on discussion with Eli, but pending
further review.
Also, renamed `#:first' to `#:before-first' and `#:last' to
`#:after-last' to be more clear, more parallel ro `#:before-last',
and avoid a collision with prominent function names.
The generics library now generates a `name/c` macro
for a generic interface `name`. The combinator can be
used to contract instances (or constructors) of a
generic interface across standard contract boundaries.
A arity-reduced procedure-valued `prop:procedure' was not handled
correctly, for example.
A good candidate for random testing? I had the right kind of test in
place, but only for an arity of 1. It turns out that testing any other
arity would have exposed the problem, so now there are tests with
arity 0. If I had randomly generated procedures instead of manually
constructing `f0' through `f1:+' in "procs.rktl", then maybe I would
have more naturally generalized the arity testing, too. Then again,
I did already have relevant inputs, and it was the testing of inputs
that was too specific.
Closes PR 12870
