| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
OTP-14527
|
|
All operations will now yield appropriately, allowing them to be used
freely in concurrent applications. This commit also deprecates the
functions listed below, although they won't raise deprecation
warnings until OTP 21:
zlib:adler32
zlib:crc32
zlib:inflateChunk
zlib:getBufSize
zlib:setBufSize
The behavior of throwing an error when a dictionary is required for
decompression has also been deprecated.
|
|
|
|
* john/erts/fix-port-leak/OTP-13939/ERL-193:
Add a testcase for OTP-13939/ERL-193
Mark socket disconnected on tcp_send_or_shutdown_error
# Conflicts:
# lib/kernel/test/gen_tcp_misc_SUITE.erl
|
|
|
|
|
|
|
|
This reverts commit 5d9bb41114544c9205a8b8f26642bad8231e8d4e.
|
|
'raimo/kernel/inet_ntoa_1-IPv6-lowercase/ERIERL-20/ERL-429/OTP-13006'
* raimo/kernel/inet_ntoa_1-IPv6-lowercase/ERIERL-20/ERL-429/OTP-13006:
Accept IPv6 address %suffixes when parsing
Adjust inet:ntoa/1 to RFC5952, but not deeper
Update inet:ntoa according to modern RFCs
|
|
This only implements decimal suffixes, and furthermore only with
a leading "0". A complete implementation would probably allow one
and probably two digit suffixes without.
But what primarily is missing is translating interface names to
and from interface indexes.
This also only implements the parsing, and uses the FreeBSD trick of
squeezing in the Scope Id in the second 16-bit word of the fe80::/64
or ff02:/64 address prefix. But inet_drv is not prepared to handle
this word, so it might only work on FreeBSD, not being supported even
there... So inet_drv needs to handle this too.
|
|
There are a multitude of RFC:s that point to each other and some of them
suggest different addresses with IPv4 suffixes. Use the IPv4 suffix
text representation only for the old well known ::a.b.c.d and
::ffff:a.b.c.d prefixes. The others seems to be moving targets.
|
|
* lukas/erts/sendfile_sunos_fix/OTP-14424:
erts: Fix sendfile closeduring scenario on sunos
|
|
On Solaris, giving a too long sfv_len results in an
EINVAL error, but data is still transmitted and len is
correctly. So we translate this to a success with that
amount of data sent. This may hide some other errors
that causes EINVAL, but it is the best we can do for now.
|
|
* lukas/erts/20_minor_fixes:
kernel: Iterate for correct time in os_SUITE:perf_counter
|
|
|
|
|
|
Leaving the cover server running could cause problems in other
test cases.
|
|
Because this test suite did not cleanup after itself,
the error_logger_warn_SUITE:file_utc failed to start its node
that it needed for the test.
|
|
* lukas/kernel/ipv6_skip_tc_fixes:
kernel: Skip ipv6 tcs when cpiv6 is disabled
|
|
For some reason doing a gen_udp:open(0, [inet6,{ipv6_v6only,true}])
works on some systems where ipv6 is disabled. So instead we
explicitly require the localhost ip to do the test at which seems
to work better.
ipv6 was disabled in sysctl using the line net.ipv6.conf.all.disable_ipv6=1
when this behaviour was found.
|
|
* bjorn/cuddle-with-tests:
code_SUITE: Remove unused functions
Eliminate warning for variable 'Config' being unused
compile_SUITE: Don't create a directory called 'core'
Don't crash in end_per_testcase/2 in code_SUITE:on_load_embedded/1
|
|
This option allows to suppress output of info_msgs to error_logger when
recoverable errors are encountered in disk_log by setting the value to
'true'.
Defaults to 'false', the backwards compatible behaviour.
|
|
|
|
|
|
* zandra/kernel/nodenames/OTP-13805:
kernel: Do not allow unicode in nodenames
|
|
Verify that unicode isn't used in a node name and logi an info
message if it is. Filter the head part of the name a bit harder
to work with epmd.
|
|
|
|
bind to device is needed to properly support VRF-Lite under
Linux (see [1] for details).
[1]: https://www.kernel.org/doc/Documentation/networking/vrf.txt
|
|
* lukas/erts/20_minor_fixes:
erts: Rebuild etc executables if config.h changes
erts: Fix new gcc warning in check io
kernel: Add mem check to prim_file:large_write tc
erts: Fix two compiler warnings on OS X
erts: Fix erts_debug:df function info output
erts: Get rid of some unused function warnings on os x
|
|
|
|
* sverker/remove-latin1-atom-encoding:
kernel: Fix erl_distribution_wb_SUITE:whitebox
kernel: Remove pg2_SUITE:compat
erts: Remove fun_r13_SUITE
stdlib: Remove test cases for R12 io protocol
kernel: Make DFLAG_UTF8_ATOMS mandatory
kernel: Rewrite distribution flag verification
tools: Update assumptions in lcnt about external atom format
stdlib: Tweak beam_lib_SUITE whitebox assumptions
orber: Remove hard dependency to external atom format
kernel: Try mend disk_log whitebox tests
erts: Mark latin1 atom encoding as deprecated
jinterface: Do not generate atoms on old latin1 external format
erl_interface: Do not generate atoms on old latin1 ext format
erts: Do not generate atoms on old latin1 external format
erts: Fix faulty ASSERT for failed dec_term
|
|
* bjorn/cuddle-with-tests:
Ensure code_SUITE:on_load_embedded/1 does not leak links
|
|
for OTP 20.
|
|
as it tests removed compatibility with R13 node.
|
|
after atom encoding got 1 byte smaller when changed
from
<<$d, Len:16, Bytes/binary>>
to
<<$w, Len:8, Bytes/binary>>
|
|
The application now has an own repo, https://github.com/erlang/typer
|
|
Make sure that the symlink created in the lib directory is
deleted even if the test case fails.
|
|
|
|
Running this test when for some reason stdlib has been
unstickied could cause the emulator to die. So we check first
to make sure that the expected files are sticky.
|
|
|
|
|
|
* maint:
Store messages for 'rex' and 'error_logger' off heap
file: match enoent and enotdir in path_open
|
|
Performance for processes that receive huge amounts of
messages can be increased by storing the incoming messages
outside the heap (that avoids copying the message in a
garbage collection).
Two OTP processes that are known to receive many messages
are 'rex' (used by 'rpc') and 'error_logger'.
|
|
The new chunk stores atoms encoded in UTF-8.
beam_lib has also been modified to handle the new
'utf8_atoms' attribute while the 'atoms' attribute
may be a missing chunk from now on.
The binary_to_atom/2 BIF can now encode any utf8
binary with up to 255 characters.
The list_to_atom/1 BIF can now accept codepoints
higher than 255 with up to 255 characters (thanks
to Björn Gustavsson).
|
|
* richcarl/kernel/disk_log_tweaks/OTP-14057/PR-1245:
Pass log format through from handle()
Improve caching in disk_log
Use pattern matching for records where suitable
Eliminate more code duplication
Simplify for rflat
Clarify that the type for disk log data is iodata()
Pass through known size instead of recomputing
Use iolist_size instead of local function
Only read log format once in collect loop
Eliminate some code duplication
Rename internal function for clarity
Minor documentation cleanup
|
|
Avoid starting timers for flushing when the written data is empty or larger
than the max cache size. Previously, a single huge write to an empty cache
would be put in the cache until the next write or the timer event. Also
increase the cache size from 16K to 64K.
|
|
|
|
These functions use the MD5 beam/native checksum to determine whether the
code for a module has changed on disk and is a candidate for loading.
|
