Commit Graph

40 Commits

Author SHA1 Message Date
Eli Barzilay
858e7dd94e test forbidden nested calls
svn: r13342
2009-01-31 20:24:35 +00:00
Eli Barzilay
339aa71441 even better
svn: r13341
2009-01-31 20:23:11 +00:00
Eli Barzilay
3956dc1530 more improvements to all four variants of exits
svn: r13340
2009-01-31 20:06:36 +00:00
Eli Barzilay
4b7e4e4322 misc improvements, add test for (exit)
svn: r13339
2009-01-31 19:46:33 +00:00
Eli Barzilay
67471cc62b cheap hack to find which memory error test is causing problems
svn: r13335
2009-01-31 15:48:31 +00:00
Eli Barzilay
43124c2ec5 more adjustments
svn: r12863
2008-12-15 21:37:55 +00:00
Eli Barzilay
f8dff60a01 adjust limits on tight test, others two, clarified comments
svn: r12862
2008-12-15 21:32:50 +00:00
Matthew Flatt
25b27a8b08 more memory-accounting repairs, so that sandbox tests now pass
svn: r12861
2008-12-15 21:15:24 +00:00
Eli Barzilay
0786da10cb svn: r12860 2008-12-15 20:43:28 +00:00
Eli Barzilay
04d0b55134 raise the limit for the problematic test, and compensate by doing more iterations
svn: r12855
2008-12-15 18:18:06 +00:00
Eli Barzilay
c62595772f Finally enabled using a sub code-inspector properly, with a new
'read-bytecode permission mode.  Added tests and (crappily)
documented.

svn: r12846
2008-12-15 01:07:00 +00:00
Eli Barzilay
9295a1dc27 * Fix memory test
* Better organization of `user-eval'
* Deal with cases when the evaluator dies outside of our control

svn: r12841
2008-12-13 23:26:44 +00:00
Eli Barzilay
39a69b30fd Don't set memory limits or test for them when there is no memory
accounting.

svn: r12839
2008-12-13 22:44:02 +00:00
Eli Barzilay
813eb20e35 termination message now indicates suicides too (both killing the
thread or shutting the custodian)

svn: r12838
2008-12-13 22:39:49 +00:00
Eli Barzilay
188489b34f * Added `exn:fail:sandbox-terminated' for all sandbox termination
kinds.
* The default exit handler kills the sandbox, with an appropriate
  termination message.
* Also, sending an eof to the sandbox shows 'eof as the reason
* Added some tests for these.
* Still need to deal with termination messages for evaluator suicide.

svn: r12837
2008-12-13 22:12:16 +00:00
Eli Barzilay
55f1df58e1 Make sure that the module is also instantiated under the evaluation
limits too (the instantiation was happening outside of the limited
thunk)

svn: r12830
2008-12-13 16:41:07 +00:00
Eli Barzilay
3fcc58c05e minor improvements
svn: r12789
2008-12-12 13:43:05 +00:00
Eli Barzilay
a1222d66ca * Implementer global sandbox memory limit and added
`sandbox-memory-limit' to set it
* Added `evaluator-alive?'
* Undo accidental commenting of most tests

svn: r12786
2008-12-12 12:45:08 +00:00
Eli Barzilay
a0d1baea00 When an expression evaluation dies out of memory, the sandbox stays alive
svn: r12785
2008-12-12 11:22:08 +00:00
Eli Barzilay
eb54b1a5b2 * Added `call-in-nested-thread*', and used it to improve
`call-with-limits'; specifically, it can now distinguish killing the
  main thread or shutting down the custodian from an out-of-memory
  error.
* This also makes it possible to have threads created and stay alive
  between evaluations.
* Added tests for this.

svn: r12673
2008-12-02 12:23:35 +00:00
Eli Barzilay
e0917e1a5e * Added `call-in-sandbox-context'.
* Removed `get-namespace' since it's generalized in the above.
* Deal with killing the current thread (or the current custodian) in
  the sandbox code.  (Works only without per-expression limits --
  still need to find a solution for `call-with-limits'.)
* Added tests for this.

svn: r12670
2008-12-02 05:26:45 +00:00
Eli Barzilay
ab817097b4 * Create string/bytes output port in the user's custodian (at least
try to do so...).
* Add tests for output collection under resources -- some are
  commented since they're not working right, either a bug in this code
  that I don't see, or some problem with memory accounting.
* Clarify in the docs that output collection is still under resource
  limits.

svn: r12667
2008-12-01 22:19:58 +00:00
Eli Barzilay
30adf7980a switch to scheme/base for the sandbox interface and for use in tests
svn: r12656
2008-12-01 03:37:02 +00:00
Eli Barzilay
fc1b9cdf9d fix expected error messages
svn: r12429
2008-11-13 17:18:50 +00:00
Eli Barzilay
7d50e61c7f * Newlines at EOFs
* Another big chunk of v4-require-isms
* Allow `#lang framework/keybinding-lang' for keybinding files
* Move hierlist sources into "mrlib/hierlist", leave stub behind

svn: r10689
2008-07-09 07:18:06 +00:00
Eli Barzilay
87bb5ee526 use proper tmp directory, remove temp file
svn: r10254
2008-06-13 17:06:05 +00:00
Eli Barzilay
b1a08edd5a use new require specs in many places
svn: r8774
2008-02-23 09:42:03 +00:00
Eli Barzilay
d2f12f294c fix expected error message on redefinitions
svn: r7954
2007-12-11 11:57:35 +00:00
Matthew Flatt
8138758ea5 fix some tests
svn: r7847
2007-11-27 17:58:59 +00:00
Eli Barzilay
6c13e101bf fix bad input to raise an error as usual, added tests
svn: r7445
2007-10-07 18:12:11 +00:00
Eli Barzilay
eb72a0dd93 do not trust require spec from user code
svn: r7343
2007-09-15 07:09:06 +00:00
Eli Barzilay
392f99b717 new thread breaking in sandbox, improved with-limits to use call-in-nested-thread
svn: r7123
2007-08-19 23:27:14 +00:00
Eli Barzilay
e374d7a051 more memory!
svn: r6141
2007-05-04 09:43:58 +00:00
Eli Barzilay
e4b7d760d5 more memory for limited memory test
svn: r6119
2007-05-02 09:31:39 +00:00
Eli Barzilay
30a27b1d8b better sandbox input
svn: r5905
2007-04-09 11:11:22 +00:00
Eli Barzilay
c6f377aca8 better output specs and implementation
svn: r5903
2007-04-09 09:15:25 +00:00
Eli Barzilay
4cf75fbb65 some filename tweaks for windows
svn: r5891
2007-04-08 21:39:12 +00:00
Eli Barzilay
874c0331c3 do what you can when not in 3m
svn: r5878
2007-04-07 02:23:36 +00:00
Eli Barzilay
6aa7b2c87d typo
svn: r5877
2007-04-07 01:44:37 +00:00
Eli Barzilay
433c9a57ec new sandbox in mzlib
svn: r5873
2007-04-06 08:56:23 +00:00