Age | Commit message (Collapse) | Author |
|
The undefined_functions/1 test case will fail if it is run
using an uninstalled Erlang/OTP system (i.e. in a source
tree/git repository).
Since the test_server application is now a part of Erlang/OTP
(it was not originally), the system-wide test case that looks
for use of undefined functions will catch undefined functions.
Therefore the easiest solution is to remove the test case.
|
|
* bjorn/common_test/dialyzer-fixes:
ct_master_logs: Don't use io:format/3 with an empty variable list
common_test: Don't use undocumented features of io:format()
|
|
* egil/r16/strengthen-buffer-copies:
Replace sprintf with erts_snprintf in beam
Replace sprintf with erts_snprintf in epmd
Replace sprintf with erts_snprintf in inet_gethost
|
|
|
|
* egil/fix-configure-openbsd/OTP-10395:
Fix linking of shared libraries on OpenBSD
|
|
|
|
* egil/fix-missing-modules-app/OTP-10439:
tools: Add missing lcnt module to .app.src
percept: Add missing modules to .app.src
|
|
* siri/common_test/undefined-snmp-funcs/OTP-10088:
[common_test] Moved ct_snmp_SUITE into data dir and run as other ct tests
[common_test] Updated (minimized) config for ct_snmp test after fixing bugs
[common_test] Use ct_snmp 'agent_vsns' config value in snmp app config
[common_test] Change NotifType to atom instead of string
[common_test] Remove calls to undefined functions from ct_snmp
[common_test] Add minor test suite for ct_snmp
Conflicts:
lib/common_test/test/Makefile
|
|
* siri/reltool/misc-bugs/OTP-10012:
[reltool] Use application name from .app file instead of directory name
[reltool] Keep order from rel spec when sorting used and included applications
[reltool] Automatically add included applications in rel file
[reltool] Make reltool:install/2 work on windows
[reltool] Allow incl_cond=derived on module level
|
|
|
|
Bugs reported in OTP-10432 allowed the config for ct_snmp test to
specify only one version for agent and manager.
|
|
OTP-10432
Config parameter 'agent_vsns' in ct_snmp was ONLY used in call to
snmp_config:write_agent_snmp_files. There it is only used for deciding
if usm config shall be written (if v3 is included) or not.
The value of 'agent_vsns' was NOT added to the snmp application's
agent configuration, so the snmp application's own default value would
be used - this is [v1,v2,v3]. The result was that if v3 was not
included in 'agent_vsns', then usm.conf was not written, but when
starting the agent it would complain that this file did not exist
since snmp's default versions are all [v1,v2,v3].
This has been corrected - the 'agent_vsns' value is now inserted in
the snmp application agent configuration as {versions,AgentVsns}.
|
|
OTP-10432
In call to snmp_config:write_agent_snmp_files, the NotifType argument
was earlier a string. This has been changed to an atom since it failed
in snmp application and notify�conf was not created.
|
|
In the first traversal of library directories, reltool used only the
directory names in order to figure out application names. This would
succeed if the directory name was AppName only or AppName-AppVsn and
AppVsn consisted of integers separated by dots only. If the AppVsn has
any other format, then reltool would not find the correct application
name.
With this commit, reltool will first look for a .app file and use the
.app file name as the application name. This will allow different
formats of the version identifier in the directory name. Note that
reltool can still not sort (and select the latest) amongst version
identifiers of other format than integers separated by dots.
|
|
|
|
|
|
* bjorn/ct-fix-silent-death/OTP-9769:
Teach test_server to report severe errors to common_test
test_server_ctrl: Present "cannot create log dir" errors more neatly
test_server_ctrl: Don't die in stop_extra_tools/1
|
|
* jv/filelib_recursion_wildcard:
Allow ** in filelib:wildcard
OTP-10431
|
|
* bjorn/ts-fixes:
ts: Remove the non-working ts:clean/0,1 commands
ts: Remove the obsolete ts:index/0 command
ts: Remove the unused module ts_selftest
ts_run: Make errors from ct:run_test/1 visible
|
|
The commands no longer work now that ts is a wrapper for
common_test (instead of for test_server), and no one seems to
have missed them.
|
|
The ts:index/0 command only works with test_server log directories.
The functionality to generate an overview of all tests run is now
provided by common_test itself.
|
|
|
|
The ts wrapper module invokes common_test by calling the
ct:run_test/1 function using "erl -eval Cmd". Since
ct:run_test/1 is intended to by called from the Erlang shell,
it returns error values rather than printing them. Therefore,
when an error occurs, users of ts may not see any error
indication.
Introduce the ts_run:ct_run_test/2 wrapper which invokes
ct:run_test/1 and prints out any error that occurs.
|
|
It is bad practice to use an arbitrary string as a format
string for io:format(), since it could contain a '~' character
which could trigger a badarg exception. Therefore, replace all
io:format/3 calls that looks like:
io:format(Fd, String, [])
with:
io:put_chars(Fd, String)
|
|
The Format string argument for io:format() is not documented
to accept an iolist, so we should not depend on it.
Also, it is bad practice to use an arbitrary string as a format
string for io:format(), since it could contain a '~' character
which could trigger a badarg exception.
Fix both problems at the same time by using io:put_chars() to
display the iolist.
|
|
* henrik/undef-fix/OTP-10428:
ignore calls to gs* modules if it is not built
|
|
|
|
* bjorn/kernel/global-resolve-function-fix/OTP-10419:
global: Allow "tuple funs" as resolve functions
|
|
If a severe error occurs in test_server (e.g. failing to write to
log files), test_server would terminate without commont_test knowing
about it. Since ct_run can now return an exit code, it is important
that common_test is aware of severe problem so that it can indicate
that an error has occurred.
|
|
|
|
If it is not possible for test_server_ctrl:start_log_file/0 to
create the log file directory or log files, then stop_extra_tools/1
(called from init_tester/10) is also likely to crash and hide the
original error reason. That will be confusing to the user.
Therefore, catch the call to stop_extra_tools/1.
|
|
Commit de7e01c958ff7c9e6da4034a53567a30a4ae5792 removed support for
tuple funs. To still allow resolve functions for
global:{re_}register_name/3 to be specified as {M,F}, we'll need to
translate {M,F} to an external fun.
Also update the documentation to mention that the use of {M,F} to
specify a resolve function is deprecated, and to add a recommendation
to use an external fun.
|
|
The rel specification now dictates the order in which included and
used applications are loaded/started by the boot file. If the
applications are not specified in the rel spec, then the order from
the .app file is used. This was a bug earlier reported on systools,
and is now also implemented in reltool.
Example:
If a .app file specified
{applications,[x,y]}
{included_applications,[b,c]}
And the reltool.config has
{rel, "myrel", "1.0", [a,y,x,c,b]}
Then the boot file will load/start y before x and c before b. Earlier
x would always be started before y and b always before c due to the
order in the .app file.
|
|
Two adjacent * used as a single pattern will match
all files and zero or more directories and subdirectories.
|
|
* bjorn/stdlib/filelib-wildcard/OTP-6874:
Fix filelib:wildcard/2
|
|
If a 'rel' spec in the reltool config does not contain all
applications that are listed as {applications,Applications} in a .app
file, then these applications are autmatically added when creating the
.rel file.
For 'included_applications', the behaviour was not the same. I.e. if a
'rel' spec in the reltool config did not contain all applications that
are listed as {included_applications,InclApplications} in a .app file,
then reltool would fail with reason "Undefined applications" when
creating the .rel file. This has been corrected, so both
'applications' and 'included_applications' are now automatically added
if not already in the 'rel' spec.
I.e. for an application 'z', if z.app contains
{applications,[stdlib,kernel,x]}
{included_applications,[y]}
then reltool configuration
{rel, "myrel", "1.0", [z]}
will cause the same .rel file as
{rel, "myrel", "1.0", [stdlib,kernel,x,y,z]}
|
|
* siri/dont-allow-modvsn-in-app-file/OTP-10417:
Don't allow {Mod,Vsn} in modules list in .app file
|
|
It was possible to insert {Mod::atom(),Vsn::term()} instead of
Mod::atom() in the modules list in a .app file. This was not visible
in the documentation of .app files, but it was visible in the
documentation of application:load/[1,2] (where the .app file term can
be used directly as first argument).
The Vsn part was never used, so this possibility has now been removed.
|
|
* sverk/gcc-atomic-config-bug:
erts: Fix gcc atomic bug in ethread.h
OTP-10418
|
|
|
|
|
|
* siri/sasl/dont_destroy_test_dir/OTP-10394:
Skip release_handler test related to symlinks on windows
|
|
The erl.ini file generated by reltool:install/2 faulty pointed out
$ROOT/bin as Bindir. This is now changed to $ROOT/erts-Vsn/bin. Also,
backslashes in erl.ini needed to be escaped.
|
|
filelib:wildcard("some/relative/path/*.beam", Path) would fail to
match any file. That is, filelib:wildcard/2 would not work if the
first component of the pattern did not contain any wildcard
characters.
Noticed-by: Samuel Rivas
|
|
|
|
* siri/sasl/dont_destroy_test_dir/OTP-10394:
Make sure release_handler_SUITE does not mess up sasl test directory
|
|
According to documentation it should be allowed to set
incl_cond=include|exclude|derived, but if set to derived on module
level, reltool_server would crash. This has been corrected.
|
|
* ia/public_key/ssh_decode-comment-issue/OTP-9361:
public_key: ssh_decode now handles comments, at the end of the line, containing withespaces correctly
|
|
* ia/public_key-missing-sha224/OTP-9362:
public_key: Add missing references to sha224 and sha384
|
|
release_handler_SUITE:otp_9864 deleted parts of the
release_handler_SUITE_data directory so the test suite could not be
executed twice without re-installation. This has been corrected.
|