| Age | Commit message (Collapse) | Author |
|---|
|
Also refactor Rizzo tests to minimize them to only test
what actually can be tested. In practice this was already
done but a lot of legacy code remained. In the past timing
happened to make it look like it was possible to test
Rizzo-Dungeon mitigation in more detail, but it was based on
false assumptions that can not be applied to streams.
|
|
It looked like two messages was received in the "wrong" order,
cuasing the request_and_reply_plain test case tro fail (on
a windows machine).
On closer inspection, the reason was that an "unconditional"
receive was made when waiting for a (disconnect) callback.
But instead we received a secondary message. So, remove
the unconditional receive and instead introduce a timeout
(10 seconds).
|
|
It seems that the megaco_config process did not start (for
an unknown reason) and that resulted in a failed test case.
But starting the config process was not what the test case
was for, so it should have resulted in a skip.
Also, since we did not check the result of the config start,
we don't know why it failed.
So, move the start and stop of the megaco_config process
to the [init|end]_per_testcase function(s). If we fail to
start, then we skip (with a useful reason).
|
|
|
|
|
|
* maint:
ssl: Fix handling of ALPN in TLS 1.3
|
|
* peterdmv/ssl/fix-tls13-alpn:
ssl: Fix handling of ALPN in TLS 1.3
|
|
|
|
* ingela/ssl/restructure-tests:
ssl: Add API suites
ssl: Covered by property tests
ssl: Correct RSP/PSK and ALPN handling
ssl: Better grouping of test
|
|
This is also an effort to group test cases better.
More such commits will follow.
|
|
|
|
The new implementation use maps iterators and are more
perfomant in banchmarks by roughly 10%. More importantly,
the iterators approach allow us to short-circuit and abort
early.
fold and filter have also been changed to use iterators.
We could simply delegate to the maps' functions, but
inlining the implementation allows us to skip a double
anonymous function dispatch.
|
|
The load test suite did not properly handle attempts to
skip test cases. If we failed to start one of the slave
nodes for whatever reason (in this case we got a timeout),
this should not result in a failed test case as this is
out of the control of the test case. As it was, we *did*
try to issue a skip, but the "load controller" did not
detect it. It instead transformed it into a regular error.
|
|
|
|
Extention handling need some fixes to work
correctly for ALPN and SSL-3.0 only client/servers
do not support extensions
|
|
Regroup testes after functionality making it easier to
test all variants (all protocols, all cert types etc).
Also extend OpenSSL interop
|
|
The functions io:columns() and io:rows() only worked correctly inside
interactive erlang shells before this fix. These functions returned
{error,enotsup} before this fix even if stdout and stdin were
connected to a terminal when they were invoked from an escript or a
program started with e.g., `erl -noshell`.
This commit fixes issue ERL-717.
|
|
Removed the "compile all" compiler directive for the mreq
megaco test suite.
This is not actually used in the "top" suite (for some reason).
We leave it like that, but still fix the various issues so that
it at least work.
|
|
Removed the "compile all" compiler directive for the load
megaco test suite.
|
|
Removed the "compile all" compiler directive for the examples
megaco test suite.
|
|
Removed the "compile all" compiler directive for the digit map
megaco test suite.
|
|
Removed the "compile all" compiler directive for the config
megaco test suite.
|
|
Removed the "compile all" compiler directive for the codec
megaco test suite.
|
|
Removed the "compile all" compiler directive for the call flow
megaco test suite.
|
|
Removed the "compile all" compiler directive for the actions
megaco test suite.
|
|
Removed the "compile all" compiler directive for the appup
megaco test suite.
|
|
Removed the "compile all" compiler directive for the app
megaco test suite.
|
|
Removed the "compile all" compiler directive for the top
megaco test suite.
|
|
|
|
Reworked the megaco test generator (megaco_test_generator) behaviour
to remove compiler warnings.
|
|
Removed all use of erlang:now() and the deprecated random module
from the test suite.
|
|
Make the mib (sub-) test suite more verbose and also
do some printout when starting slave nodes (to see
just how long time it takes).
The connect test case failed on one (possibly) slow
host during what looked lite mgc start (timeout). But
since there where no printouts during that phase, there
is no way to know what was going on.
|
|
* maint:
Fix leaking SSL node
|
|
* raimo/test-cuddling:
Fix leaking SSL node
|
|
* maint:
Updated OTP version
Prepare release
|
|
* maint-21:
Updated OTP version
Prepare release
# Conflicts:
# OTP_VERSION
# lib/ssl/doc/src/notes.xml
# lib/ssl/vsn.mk
# make/otp_version_tickets
# otp_versions.table
|
|
* maint:
Do not call dist_ctrl_put_data with empty binaries
Refine the queue code
Fix extracting 0 bytes from queue
|
|
'raimo/ssl/fix-zero-fragment-buffer-handling/ERIERL-379/OTP-15328' into maint
* raimo/ssl/fix-zero-fragment-buffer-handling/ERIERL-379/OTP-15328:
Do not call dist_ctrl_put_data with empty binaries
Refine the queue code
Fix extracting 0 bytes from queue
|
|
* john/compiler/improve-validator-type-inference/ERL-998:
beam_validator: Improve type inference on inequality
beam_validator: Remove redundant calls to infer_types
|
|
* john/erts/merge-fd-file-info/OTP-15956:
file: allow read_file_info on file descriptors
|
|
|
|
This is a more proper fix for ERIERL-348. We used to think that
we wouldn't need to update the type of a variable that's no
longer referenced by a register ("dead value"), but the attached
test case pokes a hole in that assumption.
To summarize, the result of '=:='/2 is kept alive longer than one
of its arguments, which gets pruned in a state merge leaving us
with nothing to work on when we finally compare the result. This is
fine for most operations since there's no point in (say) updating
the size of a tuple we can no longer reach, but '=:='/2 updates
the types of both arguments and we risk missing out on important
information when either of them is gone.
This commit fixes the problem by merging all values that are
*reachable* from a register, rather than just those that *exist*
in a register, ensuring that all values stay around at least as
long as they're needed.
|
|
|
|
|
|
'raimo/ssl/fix-zero-fragment-buffer-handling/ERIERL-379/OTP-15328' into maint-21
* raimo/ssl/fix-zero-fragment-buffer-handling/ERIERL-379/OTP-15328:
Do not call dist_ctrl_put_data with empty binaries
Refine the queue code
Fix extracting 0 bytes from queue
|
|
* maint:
Updated OTP version
Prepare release
# Conflicts:
# OTP_VERSION
# make/otp_version_tickets_in_merge
|
|
* maint-22:
Updated OTP version
Prepare release
# Conflicts:
# make/otp_version_tickets
|
|
Both sides need to be inferred, and we should infer as if we've
made an exact match when we know that LHS is single-valued. This
was done for select_val, but we failed to do so for is_ne_exact
et al.
|
|
|
|
* maint:
compiler: Fix compiler crash introduced by OTP-15952
|
