Age | Commit message (Collapse) | Author |
|
* nox/silent-rules-fixes/OTP-11351:
Fix two small silent rules omissions
|
|
Conflicts:
erts/vsn.mk
|
|
* msassak/empty-pattern-badarg/OTP-11350:
stdlib: extended binary_module_SUITE to assure badarg for OTP-11350
Check all pattern arguments passed to binary:matches/2
|
|
OTP-11362
* sv/system_info_max_ets:
Update preloaded
add system_info(ets_limit)
|
|
|
|
* rickard/aligned-sys_alloc-carriers_maint/OTP-11318:
Implement platform specific aligned sys_alloc and use when supported
|
|
Add the no_ok_wrapper option so that the generated M:encode/2 and
M:decode/2 functions will not wrap a successful return value in an
{ok,...} tuple. Errors will cause exceptions.
Eliminating the wrapping tuple allows simpler nesting of calls.
|
|
Use 'try' instead of 'catch', and don't match anything that
cannot actually be returned from the generated encoding code.
|
|
* bjorn/asn1/clean-up-tests: (24 commits)
No longer compile the X420 specs for plain BER
Combine the testSeqIndefinite/1 and testSetIndefinite/1 test cases
Remove asn1_wrapper
Eliminate the use of asn1_wrapper
Simplify tests of CHOICE with OPTIONAL
asn1_test_lib: Add roundtrip functions and use them everywhere
Makefile: Release Emakefile to ensure warnings are treated as errors
Don't allow unused exported functions in test case helpers
Move out specific tests from asn1_test_lib
testMegaco: Remove unused exported function msg11/0
ASN.1 tests: Don't export functions that are only locally called
Remove unused pem_performance.erl file
asn1_SUITE: Reinstate test of sub-constraint
ASN.1 tests: Remove unused choice_extension.erl
Remove unused compile() functions
asn1_test_lib: Remove unnecessary loading of a compiled ASN.1 spec
Slightly clean up testX420
asn1_SUITE: Combine most tests that use External.asn1
asn1ct_gen: Clean up process dictionary after generating
asn1ct_parser2: Clean the process dictionary after parsing
...
|
|
rickard/aligned-sys_alloc-carriers_maint/OTP-11318
Conflicts:
erts/emulator/beam/erl_alloc.c
erts/emulator/beam/erl_alloc_util.c
erts/emulator/beam/erl_alloc_util.h
|
|
erts_sys_aligned_alloc() is currently implemented using posix_memalign if
it exist, or using _aligned_malloc on Windows.
If erts_sys_aligned_alloc() exist allocators will create sys_alloc
carriers similar to how this was done pre-R16.
|
|
|
|
* fredrik/update_versions_R16B03:
Update version numbers for R16B03 development
|
|
|
|
|
|
Add system_info(ets_limit) to provide a way to retrieve the runtime's
maximum number of ETS tables. Add tests and documentation for it too.
Also repair the alphabetical order of system_info/1 argument descriptions
in the documentation and in the erlang.erl clauses. Add new preloaded
erlang.erl due to that change. Also ensure all system_info/1 clauses are
represented in the erlang.xml source documentation -- a couple had been
inadvertently dropped in previous commits when other clauses were added.
|
|
* lewellyn/nosyslog/OTP-11349:
Define LOG_ERR for systems without syslog.h
Fix syslog defines
|
|
|
|
Including an empty binary as one of multiple patterns passed to
binary:matches/2 would crash BEAM with: "Cannot reallocate
1342177280 bytes of memory (of type "tmp")". This ensures each pattern
is valid before trying to match.
|
|
We used to compile the X420 specs for both DER and PER, but
9ec0a0ba changed that to compile for DER and plain BER (probably
accidentally).
Since the X420 suite contains 99 source files and compilations
takes more than 40 seconds on my computer for one backend, we only
want to do compilations that are likely to find bugs.
Compiling with the BER backend after having compiled for the DER
backend is very unlikely to detect any bugs.
Compiling for PER (as well as DER) is slightly more likely to expose
bugs, but I don't think it is worth the extra running time for the
test suite.
|
|
They use the same ASN.1 spec.
|
|
|
|
The asn1_wrapper made some sense when encode functions
could either produce a list or a binary, depending on the
backend. Now encode functions always produce a binary.
To improve readbility of the test suites, eliminate the asn1_wrapper
functions by replacing them with calls to one of the roundtrip
functions in asn1_test_lib. When it is not possible to use
the roundtrip functions, call the module in question directly.
While at it, also remove ?line macros that are near to the
touched code and use asn1_test_lib:hex_to_bin/1 instead of
home-brewn hex conversion routines.
|
|
Test ChoOptional and ChoOptionalImplicitTag with the same test code
and data.
|
|
Even if the roundtrip functions are very simply, putting them here
makes it possible to change its behavior in one place (e.g. if we
are to change encode and decode not to wrap the result in a ok tuple
in some future release).
|
|
The generated Emakefile was not copied when releasing the test
suites, which meant that the suites were compiled with default
options (without the 'warnings_as_errors' option in the generated
Emakefile).
While at it, eliminate left-over dependecies.
|
|
To keep the test case clean, add an xref test case to ensure that
there are no unused functions in the any of the test case helper
modules. Eliminate 4 functions that were exported just so that
timer:tc/3 could be used (use timer:tc/1 with a fun instead).
|
|
asn1_test_lib is supposed to be a general-purpose helper module for
test cases. Move all specific test cases into asn1_SUITE.
|
|
Also remove records, defines, and request/4 that were only used
by msg11/0.
|
|
|
|
The module tests the performance of the OTP-PUB-KEY module in the
public_key application, so it is off-topic and should not be
reinstated.
|
|
A test of sub-constraints for PER/UPER was lost in commit 9ec0a0babace.
|
|
The testChoExtension module tests the same thing (and more) as
choice_extension.erl.
|
|
|
|
asn1ct:compile/2 calls c:c/2 to invoke the compiler and to load
the compiled code. That means that there is no need load the module
after having invoked the ASN.1 compiler.
To make sure that we'll notice if that behavior changes in the future,
extend the rtUI/1 test case to verify that the loaded module has
the correct encoding rule (if asn1ct:compile/2 would fail to reload
the module, the previous encoding rule would be returned).
While at it, it also makes sense to test calling M:bit_string_format/0
to ensure that it continues working.
|
|
|
|
By combining most tests that use External.asn,
we will only need to compile that spec once instead of 11 times.
While combining the test cases, make sure to remove the redundant code
in testPrimExternal/1 that tests primitive strings in exactly the same
way as testPrimString/1. Removing the redundant code shaves off a few
seconds of testing time.
Keep testSeqExtension/1 as a separate test case, since it is a
complicated test case that does hairy stuff to the source files.
|
|
Cleanliness.
|
|
Apart from cleanliness, the test suite runs many tests cases in
parallel, so it never hurts to reduce the memory pressure.
|
|
It seems that the original purpose is to test encoding by
directly calling the module (as opposed to calling indirectly
using asn1rt). That is tested in many other test cases (such
as testPrimStrings); thus this test case no longer serves any
useful purpose.
|
|
The per_GeneralString/1 test case compiles MULTIMEDIA-SYSTEM-CONTROL
and then does a single decoding. Save some compilation time by folding
the test into the h323test/1 test case.
|
|
|
|
The PER back-end no longer use a driver (or even a NIF),
and the test case would not fail if it didn't work because the
spawned processed were not linked.
|
|
|
|
|
|
|
|
|
|
* fredrik/sfwi-r15b03/OTP-11164:
erts: reorganization of functions to r15b03 standards
erts: bumped version number
erts: Add the +sfwi system flag
|
|
|
|
|