suzanne.soy/racket
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
38,051 Commits 17 Branches 54 Tags 278 MiB
31d82eb9a6
Commit Graph

38051 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Vincent St-Amour
5353dd1076 free-vars: add mode that also reports module-bound variables 
As a separate mode, for backwards compatibility.
 2015-12-07 10:42:12 -06:00
Vincent St-Amour
22caaad944 free-vars: nicer error on submodules 2015-12-07 10:31:12 -06:00
Matthew Flatt
715bdbb49e incremental GC: avoid already-finished finalizaton work 
Avoid extra work svaing finalization information and
re-chaining old-generation records for a major GC.
 2015-12-06 12:36:52 -05:00
Matthew Flatt
e44926fcee increemntal GC: compact when a major GC is forced early 
This change further defends against fragmentation when
incremental mode isn't working well.
 2015-12-06 12:36:52 -05:00
Matthew Flatt
b5131321d7 force non-inremental GC if excessive fragmentation 
Increemntal GC skips the old-generation compaction phase.
For most applications, that's ok, but it's possible for
fragmentation to get out of hand. Detect that situation
and fall back to non-incremental mode for one major GC.
 2015-12-06 12:36:52 -05:00
Sam Tobin-Hochstadt
a0c09c19ac Add test for 21316e3ebf. 2015-12-06 12:22:37 -05:00
Sam Tobin-Hochstadt
21316e3ebf Don't copy immutable vectors in vectorof if not needed. 
This happens only if the element contract is a flat contract.
 2015-12-06 10:45:13 -05:00
Sam Tobin-Hochstadt
b8d4248053 Copy immutable vector more efficiently. 
Thanks to @mflatt for the suggestion.
 2015-12-06 10:31:01 -05:00
Robby Findler
e814d742a7 fix the late-neg construction for flat contracts 
Thanks to Sam for finding this problem!
 2015-12-05 23:03:06 -06:00
Robby Findler
8b3369f81c set up some information for TR 2015-12-05 15:35:46 -06:00
Matthew Flatt
d87e3ead7f fix test for the case that the addon dir is missing 2015-12-05 14:41:22 -05:00
Leif Andersen
f88ba77a5c Fix travis to build against latest Racket build. 2015-12-04 18:01:57 -05:00
Matthew Flatt
fef695f066 fix call-with-atomic-output-file for pathless filename 
Closes #1156
 2015-12-03 08:05:17 -07:00
Matthew Flatt
724dc2fdbf fix namespace-mapped-symbols forcing of lazy binding info 2015-12-02 13:14:28 -07:00
Sam Tobin-Hochstadt
a6eb00a41c Avoid warning for unused variable. 2015-12-02 10:01:01 -05:00
Matthew Flatt
345b1a8187 refine the Guide section in incremental GC 2015-12-01 17:47:29 -07:00
Matthew Flatt
d56e7309ad GC: fix yet another problem counting allocated bytes 2015-12-01 16:21:32 -07:00
Jay McCarthy
67bf4349ca Revert "Assume that files that start with . are not modules" 
This reverts commit fc34292486.
 2015-12-01 14:21:09 -05:00
Jay McCarthy
cfc28ee82a Implement . ignoring directly in test, because module suffixes has other uses 2015-12-01 14:21:05 -05:00
Gustavo Massaccesi
b175241961 Flatten nested begin and begin0 forms 
The nested begin/begin0 are flattened at read time, but some
optimizations may create new instances.
 2015-12-01 13:50:16 -03:00
Matthew Flatt
7e949d5513 GC: skip memory accounting if demand goes away 
Memory accounting is enabled on demand; if demand goes
away --- as approximated by no live custodians having
a limit or previously been queried for memory use ---
then stop accounting until demand resumes.
 2015-12-01 08:14:38 -07:00
Matthew Flatt
bef34606cb incremental GC: fix handling of ephemerons 
Fix the case that an old-generation finalizer ends up
on a modified page after all old-generation marking
is complete.

Also, make sure epehemerons are checked after previous
marking that may have left the stack empty.
 2015-12-01 05:50:06 -07:00
Matthew Flatt
03302c3f30 repair an assertion in the GC 2015-12-01 05:50:06 -07:00
Matthew Flatt
8363144818 incremental GC: make finalization more incremental 2015-12-01 05:50:06 -07:00
Matthew Flatt
f30d8bd562 incremental GC: fix overcount of immobile objects 2015-12-01 05:50:06 -07:00
Matthew Flatt
d306ecdf3a incremental GC: fix accumulation of page-repair work 
Too much work was being saved for the final step of a
major GC.
 2015-12-01 05:50:06 -07:00
Matthew Flatt
e9c722cf22 re-tune incremental GC parameters 
With immobile-object allocation repaired, smaller increments
work.
 2015-12-01 05:50:06 -07:00
Matthew Flatt
7901962647 incremental GC: fix phantom-byte counting 2015-12-01 05:50:06 -07:00
Matthew Flatt
fa3cabd681 incremental GC: avoid allocating immobile on old-generation page 
The allocation strategy for immobile objects avoids some fragmentation
in non-incremental mode, but it interferes with finishing up an
incremental major collection, so trade some fragmentation for
an earlier finish (which is far more likely to use less memory
instead of more, despite extra fragmentation).
 2015-12-01 05:50:06 -07:00
Phil Nguyen
734563a7f4 remove duplicate examples in doc for arity=? and arity-includes? 2015-11-30 10:40:06 -06:00
Jay McCarthy
fc34292486 Assume that files that start with . are not modules 2015-11-30 08:30:25 -05:00
Jay McCarthy
091db74ab6 Don't include directory when using default suffixes 2015-11-30 08:29:54 -05:00
Matthew Flatt
817fdad2d5 Windows with MinGW: Fix network address resolution 
Use the same code as for MSVC compilation, which is as simple
as defining `HAVE_GETADDRINFO`.

Closes PR 15192
 2015-11-29 17:31:47 -07:00
Vincent St-Amour
e7e75c2292 Fix exponentiation of negative single-floats and moderately large bignums. 2015-11-29 15:46:56 -06:00
Matthew Flatt
a389678556 improve interaction of incremental mode and finalization 
Really, just improve when majors GCs are forced to trigger
further finalizations. This improvement makes `(collect-garbage)`
followed by `(collect-garbage 'incremental)` move more
reliably into incremental mode.
 2015-11-28 15:54:45 -07:00
Robby Findler
71d80bace5 whoops: don't want to do the precondition check twice 2015-11-28 16:27:21 -06:00
Robby Findler
8cc2e27ca7 improve error checking in version/utils 
closes #1152
 2015-11-28 16:26:09 -06:00
Robby Findler
86934d4a4f adjust contract printing so that it looks like 
the constructors in print mode and so that it
cooperates with pretty printing to get some
newlines in there
 2015-11-28 13:47:27 -06:00
Matthew Flatt
c9e9b4e400 add info on incremental GC mode to the Guide 2015-11-28 11:55:43 -07:00
Matthew Flatt
9407afa0a2 faster GC traversal of shared closure prefixes 2015-11-28 11:21:19 -07:00
Matthew Flatt
89807d178e fix accounting of nursery for logging 
The pre-GC count didn't include the obejcts allocated on
the most recent nursery page.
 2015-11-28 10:36:12 -07:00
Matthew Flatt
c3ac1c6bf4 reduce incremental work on (collect-garbage 'minor) 
When `(collect-garbage 'minor)` is combined with incremental
mode, do less incremental work. At the same time, don't
skip an incremental GC just because a major GC is ready.
 2015-11-28 10:36:12 -07:00
Matthew Flatt
e968cfbde4 fix accounting of phantom bytes for incremental GC 
Also fixes accounting when a phantom-bytes value is
treated as potentially having a backpointer.
 2015-11-28 10:35:24 -07:00
Matthew Flatt
81e0636843 fix counting of interior-pointer-allowed objects 
Words versus bytes.
 2015-11-28 09:22:11 -07:00
Matthew Flatt
ca0e9b8b2f report timing of incremental step 2015-11-28 09:22:11 -07:00
Matthew Flatt
d37bfd45ae avoid unnecessary closure-fixup work in incremental mode 2015-11-28 09:22:11 -07:00
Matthew Flatt
46fe53fadb incremental GC: avoid extra traverals of weak-link lists 
When incremental mode has already gone through weak boxes
and arrays, a major GC doesn't need to see them again.
 2015-11-28 09:22:11 -07:00
Matthew Flatt
ad2dd24fb8 incremental GC: split "incremental" finalization to single step 
Don't finalize after performing some marking work, since
finalization is in one chunk, so it's best to avoid combining
that delay with others.
 2015-11-28 06:09:39 -07:00
Matthew Flatt
493eb1de7f tweak incremental-GC parameters 
Based on experiments with a few programs.
 2015-11-27 15:08:54 -07:00
Matthew Flatt
ba8103bbde add PLT_INCREMENTAL_GC 
Although calling `(collect-garbage 'incremental)` in a program with
a periodic task is the best way to request incremental collection, it's
handy for some experiments to have an environment variable that turns
it on permanently.

This change also makes incremental-mode minor collections log as "mIn"
instead of "min", and it changes the first field of the logged
`gc-info` structure to be a mode symbol instead of a boolean.
 2015-11-27 08:02:18 -07:00