two different modules. Not sure who might be using this, so
it seems unwise to actually fix this problem.
Also document the fact that calling this function is unlikely
to be useful unless you know the package is already installed.
closes PR 12719
Doc rendering has been occasionally crashing in Pixman/Cairo. It might
have something to do with a non-atomic callback, so let's try avoiding
that, first. (If that doesn't work, the next step is to avoid using
Cairo in a multi-threaded way, for which this is a first step, too.)
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 examples make up the majority of the front page, and it turns out
that google used some of these links in "sitelinks" in search result.
These links should be the main entry points, and we ended up having such
gems as "Printf" be there. They provide a way to "demote" URLS so they
won't show there, but doing that for one link means that another will
pop in (and there's a limit of a 100 such demotions). So add this in an
attempt to make it not consider them as things that can be used in the
sitelinks. This might be reverted if it doesn't help.
(Note that this is discouraged as a way to "sculpt pagerank", but the
purpose here is not to affect how they crawl the web pages and make page
reputation flow, but only to cover up for a missing feature that ends up
with horrible sitelinks. The pages are scanned anyway from the doc
pages.)
* The main page's title appears in search results, so use "The Racket
Language" instead of "Racket". Additional minor tweaks to a few more
titles to make them presentable as search results.
* Add a `#:description' keyword to make it easy to add it to pages. Add
such texts to the main page, downloads, community, learning, people,
and mailing list pages.
* Add a <meta name="robots" content="NOODP" /> to the front page, to
avoid getting the ODP (dmoz) blurb, hopefully the new and improved
description(s) will be used instead. (If not, this should be
reverted.)
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
This tweak avoids adding extra space when the white-label font is
taller than the line it's attached to. It also makes the label
disappear in IE 6, but I think we can live with that.
See PR 12860; some of problem related to the PR were "fixed" by
adjusting the guarantees that are specified in the documentation.
Another problem was that non-consecutive bytes could be returned.
For all currently supported platforms, the result was already
portable, despite the documentation's hedging.
Also fixed up the documentation in other ways, such as the fact
that `seconds->date' returns a `date*'.
* The old function was removed completely, people will get it from
`racket/base' anyway.
* I also removed its documentation. I thought about leaving a note in,
but if `define-ffi-definer' is the preferred style, then this should
be done when there's a way to make `define-ffi-definer' use it. (Eg,
some new #:keyword that adds a way to change the defined name.)
* Note that the function is added to `racket/private/string' and not to
`racket/string' because the latter deals only with strings, and the
new function accepts byte strings too. It might be better to start a
new `racket/regexp' module for these functions.
As the documentation says, it's supposed to be ok to use the
same evt multiple times or in multiple threads, but an internal
buffer was allocated incorrectly, so that multiple/concurrent
uses could go wrong.
Closes PR 12860
Add extra intitial-message lines, use "..." on a field name
to indicate that it could reasonably be hidden by default,
and refine some existing messages.
Includes the addition of 'overflow and 'start-overflow-work
events, whcih are effectively specializations of 'sync and
'start-work to expose overflow handling.
Also, fix a bug related to a potential GC during mark-stack
restore from a lightweight continuation.
This fixes drdr, which is using a pre-submodules version of racket.
There was also a use of one-argument `string-join' that is fixed that
has been there for a while but drdr uses the api directly.
- added some color (mostly to try to disambiguate the lines)
- several of the things named '*-element' actually belong under content, not element.
- element has 'content', not the substructs.
- convertible?s are content's.
- the 'content' field in an element is not a list, but simply a content.
- there are a bunch of things under target-element.
- image-element was missing fields
- collect-element was missing the collect field
Simplified code, and additional testing for redundant properties, and
for properties that can be collapsed into a parent directory (this might
be dropped or used only for notification on differences, see
commentage).
Also, make drdr test itself reflectively.
The code is improved, and also the default changes for *.rkt files:
there is no need for any flags -- "-q" is redundant, and with no flags
it's doing "-u" which is the same as "-t" (with an added "-N").
This was broken in the class100 port. The init arguments
previously handled by the class100*/kw macro need to be
explicitly passed down.
Also, an (init-rest) is needed at the leaves of the class
hierarchy to ensure that internal super init args don't
leak via error messages. (the class100 macro always
inserts these)
Added a test file so similar breakage is detectable in the
future.
1. Lots of #lang-ization, other racketizations, code improvements, etc.
2. Some files that were not working now do.
3. "collects/tests/aligned-pasteboard" had some files that were near
duplicates of "collects/mrlib/private/aligned-pasteboard/tests".
I've removed the former since in a few places it looked like an older
version (eg, there were bogus references to a non-existent
"utils.rkt"). The former has more files that are in runnable
condition now.
4. There are plenty of tests that look like they're failing, but it
wasn't shown since they just return #f, and when they were running
with a "-f" these results weren't displayed.
5. I have no idea about the code, this is all just reshuffling and minor
editing.
There were many such properties that were leftover from before the big
GUI rewrite where the gracket executable was needed, and now there is no
longer any need for these things. There are two exceptions:
* There is a drdr feature that forces all tests that use gracket to run
sequentially, which is used for tests that rely on some actual state
of shown windows. I ran all of the files and monitored for windows
popping up, and left the test to use gracket in these cases.
* There are a few cases where running "racket -f <file>" resulted in an
unbound identifier, whereas "gracket -f <file>" does not. These are
because of the different global namespace, and I will fix them (props
and the files) in a different commit.
One of the expected results of this is that a drdr might run faster,
because the serial-running hack above should now be applied only where
relevant. (And some of the drracket tests are very long.)
This commit is pushed by itself to identify possible problems with a
drdr run that tests only this change.
Check that the entries are sorted, and check that there's a
`responsible' for all files on disk, except for known exceptions. It
might be better to use "git ls-tree HEAD <path>" instead of looking at
the filesystem, but then it won't work on drdr.
For example, a syntactic form box is labeled with "SYNTAX". Forms
such as `defform' and `defthing' now support a `#:kind' option
for setting the label.
