Age | Commit message (Collapse) | Author |
|
|
|
In 9870d22b2401b, the timetrap for large_write/1 was accidentally
reduced from the default 30 minutes to only 1 minute. That is not
enough for some older computers.
|
|
abnormal1/1 started failing sporadically when 22785cd1e21
changed using the messages_get/0 and using its own implementation:
get_messages() ->
receive
Msg -> [Msg|get_messages()]
after 1 -> []
end.
In test_server:messages_get/0, there was "after 0" instead of "after
1" as in this version.
We could change back to using "after 0", but it seems that the
test case just happened to work before the change.
Instead, read out the message that is guranteed to be in the message
queue, and *then* check that the message queue is empty.
|
|
In 8d7d5627, test_server:sleep/1 was changed to ct:sleep/1.
Change it to timer:sleep/1 to make the timing more exact
(ct:sleep/1 will call another process).
|
|
Allow an open operation to fail with {error,enoent}. That will
happen if the executable has compiled-in support for netns,
but the computer has not been configured for netns.
|
|
|
|
Some test machines have quite small hard drives. Help them by
remove temporary files in end_per_test_case/2.
|
|
On modern versions of Windows, creation of symlinks is supported,
but our test machines usually don't have permissions for creating
symlinks. Therefore, the has_links/0 that attempts to predict whether
link creation will work by testing version numbers will not do
any good.
Refactor the code so that test cases don't fail when links cannot
be created because of missing permissions.
|
|
There is never any corresponding get/1.
|
|
Seems that it can happen on Windows.
|
|
On Windows, we might not have permission to create symlinks.
Skip the test case instead of failing.
|
|
On old Solaris versions, many gen_sctp test cases fail. It is not
clear whether the implementation or the test cases or both must
be fixed.
The number of failed test cases is a problem, as other failed
test cases in the Kernel application might not be noticed.
For now, skip most test gen_sctp test cases on old Solaris platforms,
until the day that someone can mend the tests case and/or
implementation (or test machines die). Keep a few smoke test
cases.
|
|
|
|
Avoid interference between test cases.
|
|
On Windows, the pathname is likely to start with "C:/Program Files".
|
|
Eliminate gratuitous use of funs.
Don't limit the send attempts to 1000; it might not be enough.
Rely on four minut timetrap to fail the test case if there will
never be a send timeout error.
|
|
|
|
|
|
|
|
|
|
* egil/fix-filename_basedir-windows:
stdlib: Fix filename:basedir/2,3
|
|
|
|
|
|
* bjorn/optimize-start-up/OTP-13368:
Update preloaded modules
init: Load modules in parallel using the new loader BIFs
systools_make: Add commonly used modules to mandatory_modules/0
compile: Pre-load compiler modules when invoked from 'erlc'
|
|
* bjorn/asn1/modernize-tests:
Eliminate suite and doc clauses
Replace ?config() macro with proplists:get_value()
Modernize timetraps
Eliminate ?line macros
|
|
* bjorn/macosx-configure:
configure: Remove obsolete --enable-darwin-* options
|
|
|
|
|
|
* egil/fix-ttsl_drv-log:
Fix ttsl_drv logging without TERMCAP
|
|
* lukas/erts/scheduler_SUITE_compile_win/OTP-13428:
Fix dirty scheduler tc on windows
|
|
* lukas/kernel/error_logger_warn_fix/OTP-13184:
kernel: Fix progress check in error logger warn suite
|
|
* lukas/erts/fix-vsyslog-freebsd/OTP-13427:
erts: Fix run_erl syslog prototypes for freebsd
Conflicts:
erts/etc/unix/run_erl.c
|
|
* lukas/eldap/asn1hrl/OTP-13426:
eldap: Make sure to build out ASN.1 header
|
|
* lukas/erts/fix_scheduler_suspend/ERL-94/PR-978/OTP-13425:
erts: Fix install of suspend handler
Conflicts:
erts/emulator/sys/unix/erl_unix_sys.h
erts/emulator/sys/unix/sys.c
|
|
Good for test cases.
|
|
|
|
Good for test cases.
|
|
Conflicts:
lib/ssh/test/ssh_connection_SUITE.erl
|
|
|
|
|
|
|
|
|
|
Handle the case when the environment APPDATA is not set.
|
|
Remove the options:
--enable-darwin-universal
--enable-darwin-64bit
The --enable-darwin-universal option turns on universal binaries
(Intel/PPC). It does not work on modern releases of OS X, and
OTP 19 will most not likely build on an OS X release that still
supports universal builds.
The --enable-darwin-64bit option is not needed, because 64-bit builds
are default on modern OS X releases. There is also the generic
--enable-m64-build option.
|
|
=== OTP-17.5.6.9 ===
Changed Applications:
- diameter-1.9.2.4
- erts-6.4.1.6
- ssl-6.0.1.2
Unchanged Applications:
- asn1-3.0.4
- common_test-1.10.1
- compiler-5.0.4
- cosEvent-2.1.15
- cosEventDomain-1.1.14
- cosFileTransfer-1.1.16
- cosNotification-1.1.21
- cosProperty-1.1.17
- cosTime-1.1.14
- cosTransactions-1.2.14
- crypto-3.5
- debugger-4.0.3.1
- dialyzer-2.7.4
- edoc-0.7.16
- eldap-1.1.1
- erl_docgen-0.3.7
- erl_interface-3.7.20
- et-1.5
- eunit-2.2.9
- gs-1.5.16
- hipe-3.11.3
- ic-4.3.6
- inets-5.10.9
- jinterface-1.5.12
- kernel-3.2.0.1
- megaco-3.17.3
- mnesia-4.12.5
- observer-2.0.4
- odbc-2.10.22
- orber-3.7.1
- os_mon-2.3.1
- ose-1.0.2
- otp_mibs-1.0.10
- parsetools-2.0.12
- percept-0.8.10
- public_key-0.23
- reltool-0.6.6
- runtime_tools-1.8.16.1
- sasl-2.4.1
- snmp-5.1.2
- ssh-3.2.4
- stdlib-2.4
- syntax_tools-1.6.18
- test_server-3.8.1
- tools-2.7.2
- typer-0.9.8
- webtool-0.8.10
- wx-1.3.3
- xmerl-1.3.7
Conflicts:
OTP_VERSION
erts/doc/src/notes.xml
erts/vsn.mk
lib/diameter/doc/src/notes.xml
lib/diameter/src/base/diameter_service.erl
lib/diameter/src/diameter.appup.src
lib/diameter/vsn.mk
lib/ssl/doc/src/notes.xml
lib/ssl/vsn.mk
otp_versions.table
|
|
* egil/fix-percept-now/OTP-13422:
Modernize test suites
Use monotonic_time/1 instead of now/0
|
|
|
|
Use erlang:prepare_loading/1 and erlang:finish_loading/1 to load
modules in parallel to potentially decrease start-up times.
|
|
Add commonly used modules to the list of modules in
mandatory_modules/0. Having those modules listed will potentially
help the 'init' module to load them in parallel during start-up.
|
|
Slightly speed up 'erlc' by pre-loading the modules used
by the compiler. Write a test case to ensure that the correct
set of modules are loaded.
|