The cache was keyed on `current-library-collection-paths`, but
not other parameters such as `current-library-collection-links`,
so it was too "sticky" in the case that some parameters changed.
Adjust the cache to be specific to loaded module in a namespace's
module registry.
Don't print a simple filter for a function with more than
one argument or with a filter object that refers to an outer
variable.
This matches up with type parsing.
Closes PR 14510
- add a generator for cons/c
- make flat-rec-contract not just build the predicate, meaning:
- it has a better name
- it can do random generation
- make the procedures that come from any/c random generation indicate that from their names
- make or/c with one argument just (check and) return that argument, so
(or/c (or/c number?)) has the name 'number?'
Avoid polling a file descriptor to determine whether it has bytes;
just try to read, and then go to epoll/kevent mode when available,
skipping a poll/select.
Mac OS Classic and Palm ae long since unsupported.
TCP support implies sockets (since old Mac TCP is gone).
For what it's worth, make the build work without TCP support, although
no one ever builds that way.
For both original and @2x, adjust images to use alphas for the corner,
instead of drawing with clipping, which avoids alignment problems on
an @2x display.
The faster path applies in the common case of drawing a whole
bitmap, as opposed to a section of the bitmap.
Thanks to Laurent Orseau for tracking down this performance issue.
This one of those classic n^2-loop-that-should-be-linear bug. The fix
speeds up this program (when run in DrRacket) by about 1.75x:
(for ([x (in-range 10000)])
(display "abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz\n"))
The bug is fairly old too: I introduced this bug on March 6th,
2004. (Thanks, Sam, for making that old history accessible!)
